>
The Trust Fabric : Digital Trust Outlook
Volge
Quarterly
Q1 2026 · The Digital Trust issue
Sovereign AI
The Gulf builds its own
Vital Signs
India goes digital
Licence
Defending the audit

The Trust
Fabric

Verified identity, zero-trust access and provable integrity: why boards now fund trust the way they fund uptime.

volge.com/quarterlyComplimentary editionISSN 2977-4404 · 06

Volge Quarterly is Volge's signature series, published every quarter for enterprise customers. Each edition takes one subject, sets out where it currently stands on the published record, and makes a set of dated, falsifiable judgements on where it is going, written to give a buyer something firmer than a forecast: every figure carries its source and its date, every judgement carries the date on which it resolves, and the following year's edition grades each one hit, missed or partial.

In 2027 digital trust becomes an architecture, not a posture. The leaders build identity, resilience and assurance into the stack; the laggards keep bolting it on. The architecture is being assembled faster than it is being governed.

Figures carry their basis and their date. Where published estimates disagree, the range is given and the basis named rather than averaged away, and where a widely quoted number does not survive checking it is corrected here rather than repeated. Every chart plots a figure stated in the text, and nothing is projected beyond what a named source projects.

Contents

  1. 02The year trust stops being a posture and starts being an architecture
  2. 02The state of the sector
  3. 02The market in numbers
  4. 03The aggregate, read
  5. 03The segments and their 2027 inflection
  6. 04Geography: India and the DPDP clock
  7. 04Geography: the Gulf and sovereign identity
  8. 05Geography: the world, EUDI and mDL
  9. 06The forces shaping the year
  10. 06The fraud and deepfake inflection
  11. 06Standards and dated milestones
  12. 07Agentic identity and governance
  13. 07From posture to architecture
  14. 08The non-human majority
  15. 08The provenance turn
  16. 09The post-quantum clock
  17. 09Assurance as return on investment
  18. 10The buyer's ground
  19. 10The calendar that binds you
  20. 11Banking: the authentication deadline has passed
  21. 13Telecommunications: identity at the point of connection
  22. 14Health: registration is not adoption
  23. 15Government: sovereign identity at scale, and the machine gap
  24. 16The price of being late
  25. 17What is actually published about cost
  26. 19A readiness model that is not a vendor's
  27. 20What has actually been deployed in the region
  28. 21The machine majority, and why nobody agrees how large it is
  29. 22Six jurisdictions, one Gulf
  30. 23The Gulf, country by country
  31. 23Saudi Arabia: four instruments and one unreliable number
  32. 25The United Arab Emirates: the authentication rule that has already passed
  33. 26Qatar and Bahrain: one rule each, and silence around it
  34. 28Oman and Kuwait: a usage figure and a mandate, and little else
  35. 29The six compared: one region, six different answers
  36. 32The generative fraud picture: vendor surveys standing in for public data
  37. 33What is actually being procured, and what cannot be seen
  38. 35The machine identity gap across the six schemes
  39. 36The Calls: 2027
  40. 36The ten Calls at a glance
  41. 37The Calls
  42. 38How the Calls are made
  43. 39The year by domain
  44. 39The view from each angle
  45. 39Three scenarios for 2027
  46. 40What to watch, and the record
  47. 40The watchlist
  48. 40Sources: one of two
  49. 41Sources: two of two, and glossary
Volge Quarterly is Volge's signature series, published every quarter for enterprise customers.
Volge · Panache Editions · The Digital Trust edition
Digital TrustVolge Quarterly01 •

The year trust stops being a posture and starts being an architecture

2027 is the year digital trust is designed in rather than bolted on. Identity, resilience and assurance move from the security team's backlog into the architecture of the product itself.

For two dec­ades, trust in the di­gital eco­nomy was man­aged as a pos­ture. A con­trol was added after the sys­tem was built; an iden­tity check was ap­pen­ded at the peri­meter; a fraud rule was writ­ten after the loss was booked. The pos­ture worked while the ad­versary was human and the trans­ac­tion was between two known parties. Neither con­di­tion now holds. The coun­ter­feit­ing of a human face is a five-minute ex­er­cise, and a grow­ing share of the parties to a trans­ac­tion are not human at all. The pos­ture has run out of room.

Three shifts, each doc­u­mented in this edi­tion, force the change. The first is a fraud eco­nomy that gen­er­at­ive tools have in­dus­tri­al­ised: De­loitte's Cen­ter for Fin­an­cial Ser­vices pro­jects United States gen­er­at­ive-AI-en­abled fraud losses rising from US$12.3bn in 2023 to US$40bn by 2027, a 32 per cent com­pound rate (De­loitte, May 2024). The second is the ar­rival of ma­chine iden­tity at scale: Palo Alto Net­works counts 109 ma­chine iden­tit­ies for every human in the en­ter­prise, of which about 79 are agents (2026 Iden­tity Se­cur­ity Land­scape), and most zero-trust pro­grammes still stop at the human. The third is a stand­ards cal­en­dar that, for the first time, sets hard dates: the European Union's di­gital iden­tity wal­let, the EU AI Act's trans­par­ency and high-risk tiers, and India's Di­gital Per­sonal Data Pro­tec­tion ob­lig­a­tions all bind within the fore­cast win­dow.

Trust stopped being a posture the moment the agent got its own identity.

The through-line of this edi­tion is there­fore a single claim, made plainly: in 2027 the or­gan­isa­tions that treat trust as an ar­chi­tec­tural prop­erty (iden­tity as a first-class prim­it­ive, re­si­li­ence as a de­fault, as­sur­ance as a meas­ur­able out­put) will pull away from those that con­tinue to treat it as a con­trol to be added later. The gap is not primar­ily one of spend­ing. It is one of where trust sits in the stack. This is the dis­tinc­tion the re­port re­turns to at every level: mar­kets, stand­ards, agents, and the ten Calls that fol­low.

One cau­tion frames everything after it. The ar­chi­tec­ture is being as­sembled faster than it is being gov­erned. Wal­lets are being is­sued be­fore ac­cept­ance rules are settled; agents are being gran­ted cre­den­tials

be­fore the frame­works to au­thor­ise them exist; proven­ance is being signed be­fore the law re­quires it to be read. The year's de­fin­ing risk is not that trust in­fra­struc­ture fails to ar­rive. It is that it ar­rives un­gov­erned.

The state of the sector

A stack of markets (identity, verification, access, fraud, authentication, signature, provenance and resilience) compounding at double digits toward the end of the decade, read globally and across India and the Gulf.

The market in numbers

Digital trust is not one market but a stack of them, and every layer of the stack is compounding at double digits into the end of the decade.

The table below re­con­ciles the prin­cipal mar­kets that make up the trust layer, each on its own pub­lished basis. No at­tempt is made to sum them into a single fig­ure: the seg­ments over­lap (iden­tity veri­fic­a­tion is bought in­side fraud plat­forms; multi-factor au­then­tic­a­tion is sold in­side iden­tity and ac­cess man­age­ment) and a head­line ag­greg­ate would double-count. The value of the table is struc­tural, not arith­metic. It shows that the slow­est-grow­ing layer, iden­tity and ac­cess man­age­ment, still com­pounds faster than the di­gital eco­nomy it pro­tects, and that the fast­est (deep­fake de­tec­tion) is grow­ing from al­most noth­ing to­ward a real mar­ket in­side the fore­cast win­dow.

Market layerSize (base year)ForecastCAGRSource
Digital identity$44.2bn (2025)$132.1bn (2031)20.0%MarketsandMarkets, 2025
Identity verification$14.34bn (2025)$29.32bn (2030)15.4%MarketsandMarkets, 2025
Identity & access management$25.96bn (2025)$42.61bn (2030)10.4%MarketsandMarkets, 2025
Zero-trust security$36.96bn (2024)$92.42bn (2030)16.6%Grand View Research, 2025
Fraud detection & prevention$32.0bn (2025)$65.68bn (2030)15.5%MarketsandMarkets, 2025
Multi-factor authentication$21.11bn (2025)$51.96bn (2031)16.2%Mordor Intelligence, 2025
Passkeys / FIDO$2.2bn (2025)$10.19bn (2034)18.6%Market.us, 2025
Digital signature~$6.9–13.4bn (2025)n/a~30–40%House range, 2025
Deepfake detection$114.3m (2024)$5.61bn (2034)47.6%Market.us, 2025
Cybersecurity (context)$227.6bn (2025)$351.9bn (2030)9.1%MarketsandMarkets, 2025
Digital TrustVolge Quarterly02 •

Each mar­ket is stated on its own pub­lished scope and base year; the lay­ers over­lap and are not ad­dit­ive. Di­gital sig­na­ture is shown as a house range be­cause pub­lished siz­ings di­verge by roughly 2x on scope (e-sig­na­ture soft­ware versus the wider di­gital-sig­na­ture-plus-trust-ser­vices mar­ket).

Divergence to note

The di­gital-iden­tity mar­ket is the clearest case of houses dis­agree­ing. MarketsandMarkets puts the 2025–31 rate at 20.0 per cent; Grand View Re­search sizes the mar­ket lower ($47bn in 2025) and slower, at 13.2 per cent to 2030; For­tune Busi­ness In­sights sits between them at 19.9 per cent. The three are not re­con­ciled here. They are shown to­gether be­cause the spread (roughly seven points of CAGR) is it­self the find­ing: ana­lysts agree the layer is grow­ing fast and dis­agree on how fast, which is the nor­mal sig­na­ture of a mar­ket still de­fin­ing its own bound­ary.

The aggregate, read

Read together, the stack tells a consistent story: the trust layer is growing roughly twice as fast as the cybersecurity market that contains it, and the growth is migrating from the perimeter to identity.

Three read­ings fol­low from the table on the pre­vi­ous page. First, the centre of grav­ity has moved. Iden­tity-cent­ric mar­kets (di­gital iden­tity, veri­fic­a­tion, IAM, zero-trust) now com­pound faster than the net­work-and-en­d­point spend they are dis­pla­cing, which is the mar­ket's way of con­firm­ing that the peri­meter has dis­solved and iden­tity has be­come the con­trol plane. Second, au­then­tic­a­tion is being re­built rather than ex­ten­ded: pass­words are giv­ing way to pass­keys, and one-time pass­words to phish­ing-res­ist­ant factors, which is why the pass­key/FIDO and MFA lines grow while the un­der­ly­ing ac­count base does not. Third, en­tirely new lay­ers are ap­pear­ing where none ex­is­ted: deep­fake de­tec­tion did not exist as a mar­ket five years ago and is fore­cast at a 47.6 per cent com­pound rate to 2034 (Mar­ket.us, 2025).

  • 20.0%: Di­gital iden­tity CAGR. 2025–31, MarketsandMarkets; houses di­verge to 13.2% (GVR)
  • $247.5bn: In­form­a­tion-se­cur­ity spend, 2026f. Gart­ner; from $213bn in 2025, it­self +10.4% on 2024
  • 47.6%: Deep­fake-de­tec­tion CAGR. 2024–34, Mar­ket.us; fast­est layer in the stack
  • 16.6%: Zero-trust CAGR. 2024–30, Grand View Re­search; to $92.4bn
  • $10.5tn: Cost of cy­ber­crime, 2025. Cy­ber­se­cur­ity Ven­tures; to $12.2tn by 2031
  • 18.6%: Pass­key / FIDO CAGR. 2025–34, Mar­ket.us; the au­then­tic­a­tion re­build

The con­text line mat­ters. Gart­ner's in­form­a­tion-se­cur­ity fore­cast (a spend meas­ure rather than a mar­ket-siz­ing) reached US$213bn in 2025, up 10.4 per cent on 2024, and is pro­jec­ted at roughly US$247.5bn for 2026, with 2027 growth of 11.2 per cent in con­stant

cur­rency and US$372.8bn by 2030 (Gart­ner, 3Q26 fore­cast). That is the budget en­vel­ope. The trust-layer mar­kets above are grow­ing faster than the en­vel­ope, which means their share of the se­cur­ity wal­let is rising: buy­ers are real­loc­at­ing from peri­meter de­fence to­ward iden­tity, veri­fic­a­tion and re­si­li­ence. India's slice of that en­vel­ope is put at US$3.44bn and the Middle East and North Africa at roughly US$4bn for 2026 (Gart­ner, 2025).

The perimeter has dissolved; identity has become the control plane.

Divergence to note: total cybersecurity

The size of the mar­ket that con­tains the trust layer is it­self con­tested. MarketsandMarkets sizes global cy­ber­se­cur­ity at US$227.6bn in 2025 grow­ing to US$351.9bn by 2030 (9.1 per cent); Gart­ner's in­form­a­tion-se­cur­ity spend meas­ure is lower in ab­so­lute terms be­cause it ex­cludes parts of the hard­ware and ser­vices base; Mordor In­tel­li­gence sizes the Middle East alone at US$20.55bn in 2025 rising to US$46.39bn by 2031 (14.55 per cent). The re­gional rates run sev­eral points ahead of the global rate, which is the more dur­able sig­nal: emer­ging mar­kets are build­ing the trust layer at speed, not merely main­tain­ing it.

The segments and their 2027 inflection

Every layer of the stack has a distinct inflection point in 2027; read together they describe a single re-architecture of trust from the perimeter inward.

The tax­onomy below is the one this re­port uses through­out. Each seg­ment is defined by the ques­tion it an­swers (who is this, may they act, is this real, can it be proved) and each has a spe­cific reason its tra­ject­ory bends in 2027.

Identity & verification

Di­gital iden­tity is the found­a­tion layer, and its 2027 in­flec­tion is reg­u­lat­ory: the European wal­let moves from being is­sued to being re­quired (see Part II), turn­ing iden­tity from a product a re­ly­ing party chooses into in­fra­struc­ture it must ac­cept. Iden­tity veri­fic­a­tion (IDV), the one-time proof­ing of a new user, in­flects on the fraud side, as doc­u­ment for­gery driven by gen­er­at­ive tools (En­trust re­cor­ded di­gital doc­u­ment for­ger­ies up 244 per cent on 2023, and by 2025 put di­gital for­ger­ies at 35 per cent of all doc­u­ment fraud) forces a shift from doc­u­ment checks to bio­met­ric and proven­ance-backed proof­ing. Bio­met­rics be­comes the con­tested ground of that shift: the same gen­er­at­ive cap­ab­il­ity that de­feats doc­u­ment checks also at­tacks face-match, mak­ing live­ness and in­jec­tion-at­tack de­tec­tion the real product.

Digital TrustVolge Quarterly03 •

Access & authentication

Iden­tity and ac­cess man­age­ment (IAM) is the slow­est-grow­ing layer (10.4 per cent, MarketsandMarkets 2025) pre­cisely be­cause it is the most ma­ture; its 2027 in­flec­tion is the ex­ten­sion of ac­cess con­trol to non-human iden­tit­ies, which the in­cum­bents are only be­gin­ning to price. Zero-trust reaches op­er­a­tional ma­tur­ity as NIST's im­ple­ment­a­tion guid­ance (SP 1800-35, fi­nal­ised 10 June 2025) gives buy­ers a ref­er­ence ar­chi­tec­ture. Multi-factor au­then­tic­a­tion and pass­keys/FIDO to­gether mark the end of the pass­word: the in­flec­tion is the point at which at least one top-tier con­sumer plat­form makes pass­word­less the de­fault rather than an op­tion.

Assurance & resilience

Fraud de­tec­tion in­flects as gen­er­at­ive fraud be­comes a board-level line item rather than an op­er­a­tional loss. PKI and di­gital sig­na­ture face a slower but deeper in­flec­tion (the post-quantum mi­gra­tion) as NIST's stand­ard­ised al­gorithms begin to ap­pear in real cer­ti­fic­ate chains. Deep­fake de­tec­tion and con­tent proven­ance (C2PA) are the new­est lay­ers, mov­ing in 2027 from pilot to de­fault at the first plat­form and cam­era lines. Cyber re­si­li­ence is the wrap­per: the as­sump­tion that trust con­trols will be breached, and the meas­ure of how fast the or­gan­isa­tion re­cov­ers.

The 2027 pattern

Across all twelve seg­ments the in­flec­tion is the same in shape: a con­trol that was op­tional and ex­ternal be­comes man­dat­ory and in­ternal. That is the ar­chi­tec­tural turn stated as a mar­ket fact.

Geography: India and the DPDP clock

India has already built the identity rails the rest of the world is now standardising; its 2027 story is governance, not infrastructure, and the governance clock strikes in May.

India is the world's largest work­ing ex­ample of di­gital pub­lic in­fra­struc­ture. Aadhaar provides a na­tional di­gital iden­tity to more than a bil­lion res­id­ents; DigiLocker is­sues and holds veri­fi­able gov­ern­ment cre­den­tials; the two to­gether already do, at pop­u­la­tion scale, what the European wal­let is being built to do. The In­dian di­gital-iden­tity mar­ket is sized at US$2.18bn in 2025 rising to US$9.55bn by 2033, a 19.2 per cent com­pound rate (Grand View Re­search, 2026): growth that comes not from build­ing the rails, which exist, but from the ap­plic­a­tions, veri­fic­a­tion ser­vices and con­sent in­fra­struc­ture layered on top.

The de­fin­ing event of the year is reg­u­lat­ory. The Di­gital Per­sonal Data Pro­tec­tion Act, 2023 moves from stat­ute to en­force­able ob­lig­a­tion: sub­stant­ive du­ties (con­sent, pur­pose lim­it­a­tion, breach no­ti­fic­a­tion, and the heightened ob­lig­a­tions of Sig­ni­fic­ant Data Fi­du­ciar­ies) are set to bite from 13 May 2027. For the trust layer this is a de­mand event. Con­sent man­age­ment, iden­tity and audit be­come a live com­pli­ance mar­ket rather than a com­pli­ance as­pir­a­tion, and every or­gan­isa­tion pro­cessing In­dian per­sonal data at scale must be able to prove, not merely as­sert, that its iden­tity and con­sent flows are law­ful.

  • $2.18B: India di­gital iden­tity, 2025. To $9.55B by 2033, 19.2% CAGR (GVR, 2026)
  • 13 May 2027: DPDP ob­lig­a­tions bite. Sig­ni­fic­ant Data Fi­du­ciary du­ties be­come en­force­able
  • $8.58B: India cy­ber­se­cur­ity, 2025. To $16.86B by 2030, 14.5% CAGR (MarketsandMarkets, 2025)

The In­dian pat­tern is the in­verse of the West­ern one. In Europe and the United States the in­fra­struc­ture is being built ahead of the gov­ernance; in India the gov­ernance is ar­riv­ing after the in­fra­struc­ture. The risk pro­files dif­fer ac­cord­ingly. The West­ern risk is un­gov­erned ar­chi­tec­ture; the In­dian risk is a com­pli­ance mar­ket form­ing faster than the sup­ply of people and tool­ing to serve it. Both are vari­ants of the edi­tion's cent­ral cau­tion (that trust in­fra­struc­ture is being as­sembled faster than it is being gov­erned), and India is where that cau­tion is nearest to being tested at scale.

In India the rails were built first and the rules are arriving second; in Europe it is the reverse.

Geography: the Gulf and sovereign identity

The Gulf states have made national digital identity an instrument of sovereignty, and in 2027 the question is whether the sovereign rails extend from citizens to agents and enterprises.

The Gulf's model is neither the In­dian pub­lic-in­fra­struc­ture model nor the European wal­let-and-ac­cept­ance model. It is a sov­er­eign one: na­tional iden­tity is op­er­ated by the state as stra­tegic in­fra­struc­ture, tightly coupled to cit­izen ser­vices and to na­tional cyber-re­si­li­ence reg­u­la­tion. Saudi Ar­a­bia's Na­fath, the single-sign-on layer above the Ab­sher plat­form, au­then­tic­ates ac­cess to gov­ern­ment and, in­creas­ingly, private ser­vices for a user base numbered around 21.9 mil­lion; the United Arab Emir­ates op­er­ates UAE Pass as its na­tional di­gital iden­tity; Qatar and the wider Gulf

Digital TrustVolge Quarterly04 •

Co­oper­a­tion Coun­cil op­er­ate com­par­able na­tional schemes, each backed by a na­tional pub­lic-key in­fra­struc­ture.

Two fea­tures dis­tin­guish the Gulf tra­ject­ory. The first is that re­si­li­ence is reg­u­lated cent­rally: the Saudi Na­tional Cy­ber­se­cur­ity Au­thor­ity, the UAE's cyber-se­cur­ity reg­u­lat­ors and Qatar's Na­tional In­form­a­tion As­sur­ance frame­work set bind­ing con­trols that make trust a com­pli­ance prop­erty, not merely a com­mer­cial one. The Middle East in­form­a­tion-se­cur­ity en­vel­ope is put at roughly US$4bn for 2026 (Gart­ner, 2025), with the wider Middle East cy­ber­se­cur­ity mar­ket sized at US$20.55bn in 2025 rising to US$46.39bn by 2031, a 14.55 per cent com­pound rate (Mordor In­tel­li­gence, 2025): growth run­ning sev­eral points ahead of the global rate. The second is am­bi­tion: the sov­er­eign schemes are can­did­ates to ex­tend bey­ond cit­izen au­then­tic­a­tion into en­ter­prise and, plaus­ibly, agent au­thor­isa­tion, which is the sub­ject of one of this edi­tion's Calls.

  • ~21.9M: Na­fath users (KSA). Na­tional SSO above the Ab­sher plat­form
  • $20.55B: Middle East cy­ber­se­cur­ity, 2025. To $46.39B by 2031, 14.55% CAGR (Mordor, 2025)
  • $4B: MENA info-se­cur­ity spend, 2026. Gart­ner, 2025

The sovereign question for 2027

The Gulf schemes are ma­ture for cit­izens and im­ma­ture for ma­chines. Whether Na­fath or UAE Pass runs a cred­ible agent- or en­ter­prise-au­thor­isa­tion pilot within the year is the re­gion's clearest test of the agen­tic-iden­tity thesis, and is priced as a Me­dium-con­fid­ence Call in Part III.

Geography: the world, EUDI and mDL

The global benchmark for 2027 is Europe, where the digital identity wallet crosses from issuance to obligation, with the United States mobile driving licence and the FIDO passkey providing the two supporting datum lines.

The single most con­sequen­tial dated event in the global trust cal­en­dar is the European Union's di­gital iden­tity wal­let. Reg­u­la­tion (EU) 2024/1183, the re­form of eIDAS, entered into force on 20 May 2024. It re­quires mem­ber states to make a cer­ti­fied wal­let avail­able to cit­izens by late 2026, and (the point that mat­ters for 2027) it ob­liges re­ly­ing parties in reg­u­lated sec­tors, in­clud­ing banks, tele­com­mu­nic­a­tions, en­ergy and large plat­forms, to ac­cept the wal­let by late 2027. The tar­get is that 80 per cent of European cit­izens use a di­gital iden­tity by 2030. This is the strongest 2027 in­flec­tion in the re­port be­cause two clocks col­lide in­side the year: wal­let avail­ab­il­ity and the ac­cept­ance

ob­lig­a­tion. The mo­ment ac­cept­ance be­comes man­dat­ory, every re­ly­ing party must re-ar­chi­tect its iden­tity in­take to con­sume a wal­let cre­den­tial rather than a pro­pri­et­ary check.

The pivot, in one line

2027 is when the EU wal­let stops being is­sued and starts being re­quired. Is­su­ance is a sup­ply-side mile­stone; the ac­cept­ance duty is a de­mand-side one, and it is the de­mand side that forces re­ly­ing parties to re­build.

Two fur­ther datum lines com­plete the global pic­ture. The United States mo­bile driv­ing li­cence (mDL) is ad­van­cing through a fed­er­ated rol­lout: the Trans­port­a­tion Se­cur­ity Ad­min­is­tra­tion ac­cepts mDLs from about 22 states and ter­rit­or­ies as of Oc­to­ber 2026, Apple Wal­let sup­ports eight­een states plus Pu­erto Rico after four were added in Au­gust 2026, and Google Wal­let ten, all built on the ISO/IEC 18013-5 stand­ard, with 18013-7 ex­tend­ing present­a­tion to the open in­ter­net (pub­lished 2024–25). REAL ID en­force­ment, in force since May 2025, sharpens the in­cent­ive to carry a veri­fi­able cre­den­tial. Sep­ar­ately, the FIDO Al­li­ance's pass­key stand­ard is the mech­an­ism by which the pass­word is being re­tired; pass­key ad­op­tion at con­sumer plat­forms is the datum against which this re­port's au­then­tic­a­tion Call is graded.

The European acceptance duty turns identity from a product a relying party chooses into infrastructure it must consume.

Read across the three geo­graph­ies, the pat­tern is con­ver­gent in des­tin­a­tion and di­ver­gent in method. India builds pub­lic rails and le­gis­lates af­ter­wards; the Gulf op­er­ates sov­er­eign iden­tity as stra­tegic in­fra­struc­ture; Europe le­gis­lates ac­cept­ance and lets the mar­ket build to it. All three ar­rive, within the fore­cast win­dow, at the same ar­chi­tec­tural end state: iden­tity as in­fra­struc­ture the re­ly­ing party con­sumes rather than a check it per­forms.

$40 billion: Deloitte's 2027 forecast for United States generative-AI-enabled fraud, up from $12.3 billion in 2023. The loss is no longer an operational line. It is an architectural one.

The scale sits in­side a still lar­ger frame. The global cost of cy­ber­crime is put at US$10.5tn in 2025, rising to US$12.2tn by 2031 (Cy­ber­se­cur­ity Ven­tures, 2025). Nas­daq Ver­afin es­tim­ates il­li­cit fin­an­cial activ­ity at US$4.4tn in 2025, of which fraud losses ac­count for US$579.4bn and scams US$62bn (Nas­daq Ver­afin,

Digital TrustVolge Quarterly05 •

2025). In the United States alone the Fed­eral Trade Com­mis­sion re­por­ted con­sumer fraud losses of US$15.9bn in 2025 across about 3 mil­lion re­ports, of which in­vest­ment scams were $7.9bn (FTC, 25 March 2026), and Javelin sized iden­tity fraud and scams to­gether at US$38bn of com­bined iden­tity fraud and scams for 2025, down $9bn on the year, with tra­di­tional iden­tity fraud flat at $27.3bn (Javelin, 21 April 2026).

The forces shaping the year

Three forces move the whole field in 2027 (a generative fraud economy, a standards calendar with hard dates, and the arrival of identity for machines), and five through-lines cut across them.

The fraud and deepfake inflection

The engine beneath this edition's thesis is a single, well-evidenced fact: generative tools have industrialised fraud, turning identity attack from a craft into a commodity.

Every ar­gu­ment in this re­port ul­ti­mately rests on one meas­ured trend. De­loitte's Cen­ter for Fin­an­cial Ser­vices pro­jects that gen­er­at­ive-AI-en­abled fraud losses in the United States will rise from US$12.3bn in 2023 to US$40bn by 2027, a com­pound an­nual growth rate of 32 per cent (De­loitte, May 2024). That fig­ure is the an­chor 2027 stat­istic of the edi­tion, and it is cor­rob­or­ated from sev­eral in­de­pend­ent dir­ec­tions. En­trust's iden­tity-fraud data re­cord a deep­fake at­tempt every five minutes in 2024; by 2025 di­gital for­ger­ies were 35 per cent of doc­u­ment fraud, deep­fakes ap­peared in about one in five bio­met­ric fraud at­tempts, in­jec­tion at­tacks rose 40 per cent and deep­faked selfies 58 per cent (En­trust, Novem­ber 2025) and up 1,600 per cent since 2021, with di­gital for­ger­ies now 57 per cent of all doc­u­ment fraud (En­trust/On­fido, 2025). Sum­sub re­cords deep­fakes mul­tiply­ing ten­fold between 2022 and 2023, with re­gional in­creases of 1,740 per cent in North Amer­ica, 1,530 per cent in the Asia-Pa­cific and 450 per cent in the Middle East and Africa over the same period, and flags agen­tic-AI-en­abled scams as the emer­ging vec­tor for 2026 (Sum­sub, 2024).

  • $40B: US GenAI fraud losses, 2027f. From $12.3B in 2023, 32% CAGR (De­loitte, May 2024)
  • Every 5 min: A deep­fake at­tempt, 2024. En­trust/On­fido, 2025; for­ger­ies +244% on 2023
  • 10x: Deep­fakes, 2022–23. Sum­sub, 2024; +1,740% in North Amer­ica
A deepfake attempt every five minutes. The adversary is no longer scarce.

The struc­tural point is not the size of the loss but the change in its nature. When iden­tity at­tack was a craft, de­fence could be a pos­ture: catch the skilled at­tacker at the peri­meter. When iden­tity at­tack is a com­mod­ity (a face gen­er­ated in five minutes, a doc­u­ment forged at scale, a voice cloned from a pub­lic clip) de­fence must be ar­chi­tec­tural: proof of live­ness, proven­ance of con­tent, and con­tinu­ous veri­fic­a­tion built into the trans­ac­tion it­self. The fraud in­flec­tion is there­fore the mech­an­ism by which the whole field is pushed from pos­ture to ar­chi­tec­ture. It is the reason the rest of Part II mat­ters.

Standards and dated milestones

For the first time the trust calendar is governed by hard dates rather than aspirations, and the dates cluster inside the forecast window.

The table sets out the dated mile­stones that struc­ture 2027. What dis­tin­guishes this cal­en­dar from pre­vi­ous years is that the entries are bind­ing: reg­u­lat­ory in-force dates, stand­ards re­com­mend­a­tions and en­force­ment dead­lines, not vendor roadmaps. The clus­ter­ing is the find­ing: wal­let ac­cept­ance, AI-Act trans­par­ency, and In­dian data-pro­tec­tion ob­lig­a­tions all fall within a four­teen-month span.

InstrumentMilestoneDateSignificance
eIDAS 2.0 / EUDI WalletRelying-party acceptance duty (regulated sectors)Late 2027Reg (EU) 2024/1183, in force 20 May 2024; wallets available late 2026; 80% target by 2030
EU AI Act, Article 50Transparency duty (incl. deepfake labelling)2 Aug 2026Nearest binding date; obliges disclosure of AI-generated content
EU AI Act, high-risk tierHigh-risk obligations apply2 Dec 2027Adopted amended by the Digital Omnibus, Regulation (EU) 2026/1744, in force 27 July 2026; Annex III duties still apply from 2 December 2027 and Annex I moves to 2 August 2028
India DPDP Act, 2023Significant Data Fiduciary duties enforceable13 May 2027Consent, audit and identity become a live compliance market
NIST PQCStandardised algorithms in certificate chains2024–onwardFIPS 203/204/205 published Aug 2024; migration begins
C2PA / Content CredentialsSpec v2.2 → v2.5, published October 2026, adding asset formats and live media; CDN preservation2025–27v2.2 published 1 May 2025; Cloudflare preserving on ~20% of web (Feb 2025)
Digital TrustVolge Quarterly06 •
InstrumentMilestoneDateSignificance
W3C Verifiable Credentials 2.0Recommendation15 May 2025The data model beneath wallets and mDL
ISO/IEC mDL (18013-5 / -7)Presentation over the internet2024–25US TSA accepts 16 states/territories (2026)

Dates are in-force, ad­op­tion or re­com­mend­a­tion dates as pub­lished by the re­spect­ive bod­ies; where a phased re­gime ap­plies, the mile­stone most rel­ev­ant to 2027 is shown.

Why the cluster matters

A single dated ob­lig­a­tion can be ab­sorbed as a pro­ject. Three (wal­let ac­cept­ance, AI-Act trans­par­ency and DPDP en­force­ment) land­ing in­side four­teen months can­not. They force or­gan­isa­tions to treat iden­tity, con­tent proven­ance and con­sent as a com­mon ar­chi­tec­ture rather than three sep­ar­ate com­pli­ance ex­er­cises, which is pre­cisely the shift from pos­ture to ar­chi­tec­ture the edi­tion de­scribes.

Agentic identity and governance

The frontier of digital trust in 2027 is not the human user but the machine one; identity for AI agents is the field's fastest-moving and least-governed layer.

The en­ter­prise has quietly be­come ma­jor­ity non-human. Palo Alto Net­works counts 109 ma­chine iden­tit­ies for every human iden­tity in the or­gan­isa­tion (2026 Iden­tity Se­cur­ity Land­scape); De­linea pre­vi­ously put the ratio lower, at 46 non-human iden­tit­ies per human, but pro­jects the pop­u­la­tion to­ward 45 bil­lion and finds that 68 per cent of or­gan­isa­tions lack iden­tity con­trols for the AI and ma­chine iden­tit­ies they already run (De­linea, 2025). The two houses dis­agree on the mul­tiple and agree on the dir­ec­tion: ma­chine iden­tit­ies are mul­tiply­ing and are largely un­gov­erned. Most zero-trust pro­grammes, built for a human work­force, stop at the human.

The sober­ing coun­ter­weight is Gart­ner's fore­cast that more than 40 per cent of agen­tic-AI pro­jects will be can­celled by the end of 2027, with weak iden­tity and trust con­trols a lead­ing cause; the firm judges that only around 130 of the thou­sands of vendors de­scrib­ing them­selves as “agen­tic” are real (Gart­ner, 2025). Gart­ner also pro­jects that guard­ian agents (soft­ware that su­per­vises other agents) will con­sti­tute 10 to 15 per cent of the agen­tic mar­ket by 2030, with guard­ian-agent budgets rising from under 1 per cent of spend in 2026 to­ward 5 to 7 per cent by 2028. The fron­tier is there­fore double-edged: real enough to de­mand iden­tity ar­chi­tec­ture, im­ma­ture enough that most at­tempts to de­ploy it will fail. That com­bin­a­tion (genu­ine de­mand, high fail­ure) is the de­fin­ing tex­ture of the agen­tic year.

The stand­ards re­sponse ar­rived in the clos­ing months of 2025. The OpenID Found­a­tion pub­lished Iden­tity Man­age­ment for Agen­tic AI on 7 Oc­to­ber 2025, de­scrib­ing an “autonomy in­flec­tion point” at which autonom­ous agents break the as­sump­tions of single-sign-on and OpenID Con­nect. The Model Con­text Pro­tocol's au­thor­iz­a­tion spe­cific­a­tion, built on OAuth 2.1, was ex­ten­ded in Novem­ber 2025 to add URL-based cli­ent iden­ti­fi­ers, cli­ent-cre­den­tials flows and en­ter­prise iden­tity-pro­vider in­teg­ra­tion: the plumb­ing by which an agent can be is­sued and can present its own cre­den­tial. Okta's Cross App Ac­cess work trav­elled from the IETF OAuth work­ing group in Septem­ber 2025 into the MCP spe­cific­a­tion in Novem­ber 2025, and Okta was named a fea­tured iden­tity pro­vider for An­thropic's Claude En­ter­prise on 18 June 2026. The gov­ernance layer fol­lowed: the OWASP Top 10 for Agen­tic Ap­plic­a­tions v1.0 was re­leased on 9–10 Decem­ber 2025, enu­mer­at­ing agent-spe­cific risks from goal hi­jack­ing (ASI01) to rogue agents (ASI10), with a com­pan­ion AIVSS scor­ing frame­work: a “CVSS for agents.”

  • 109:1: Ma­chine iden­tit­ies per human. Palo Alto Net­works, 2026; De­linea put it at 46:1 in 2025
  • 68%: Lack AI iden­tity con­trols. De­linea, 2025; NHIs pro­jec­ted to­ward 45bn
  • >40%: Agen­tic pro­jects can­celled by end-2027. Gart­ner, 2025; weak trust con­trols a lead­ing cause
109 machine identities for every human, and most zero-trust programmes stop at the human.

From posture to architecture

The first through-line is the master argument of the edition: trust is ceasing to be a posture an organisation adopts and becoming an architecture it builds.

A pos­ture is added after the fact; an ar­chi­tec­ture is present from the first line. For most of the di­gital era, trust was pos­tural. Iden­tity was checked at a peri­meter, fraud was caught by rules writ­ten after losses, and se­cur­ity was a team con­sul­ted late. This worked while at­tacks were scarce and ex­pens­ive. The fraud in­flec­tion has made at­tacks abund­ant and cheap, and abund­ance de­feats pos­ture: a con­trol added at the edge can­not hold against an ad­versary that gen­er­ates a new face every five minutes (En­trust, 2025), and by 2025 ap­peared in about one in five bio­met­ric fraud at­tempts (En­trust, Novem­ber 2025). The only dur­able re­sponse is to move the con­trol in­ward: to make

Digital TrustVolge Quarterly07 •

iden­tity a first-class prim­it­ive that every trans­ac­tion con­sults, re­si­li­ence a de­fault the sys­tem as­sumes, and as­sur­ance a meas­ur­able out­put rather than an audit find­ing.

The evid­ence that this shift is un­der­way is in the mar­kets. The iden­tity-cent­ric lay­ers (di­gital iden­tity at 20 per cent, zero-trust at 16.6 per cent) are out­grow­ing the peri­meter spend they dis­place (MarketsandMarkets and Grand View Re­search, 2025). Zero-trust it­self is the ar­chi­tec­tural doc­trine made con­crete: NIST's final im­ple­ment­a­tion guid­ance (SP 1800-35, 10 June 2025) gives buy­ers a ref­er­ence design rather than a slo­gan. The stand­ards cal­en­dar re­in­forces the point by mak­ing trust prop­er­ties man­dat­ory and in­ternal: a re­ly­ing party that must ac­cept a European wal­let by late 2027 can­not bolt that on; it must re­build its in­take.

A posture is added after the fact; an architecture is present from the first line.

The op­er­a­tional test for a board is simple to state and hard to pass: can the or­gan­isa­tion prove, at any mo­ment, who and what is act­ing in its sys­tems, and can it show the proven­ance of what those act­ors pro­duce? An or­gan­isa­tion that an­swers yes has an ar­chi­tec­ture. One that an­swers “we have con­trols for that” still has a pos­ture. The gap between the two is what 2027 will price.

The non-human majority

The second through-line is demographic: the population of identities inside the enterprise is now overwhelmingly non-human, and the trust stack was built for the minority.

Every iden­tity sys­tem in pro­duc­tion was de­signed around a human being: a per­son who logs in, whose ac­cess is pro­vi­sioned by a man­ager, whose cre­den­tial can be re­voked when they leave. That as­sump­tion is now false at the level of arith­metic. Palo Alto Net­works counts 109 ma­chine iden­tit­ies for every human (2026 Iden­tity Se­cur­ity Land­scape); even De­linea's more con­ser­vat­ive 46:1 leaves hu­mans a small minor­ity, and De­linea finds 68 per cent of or­gan­isa­tions without iden­tity con­trols for their AI and ma­chine iden­tit­ies (De­linea, 2025). The ar­rival of autonom­ous agents ac­cel­er­ates the im­bal­ance, be­cause an agent is not one iden­tity but a pro­lif­er­at­ing class of them: it spawns sub-agents, holds del­eg­ated cre­den­tials, and acts across sys­tems faster than any human re­view can fol­low.

The stand­ards world has begun to re­spond. The OpenID Found­a­tion's Iden­tity Man­age­ment for Agen­tic AI (7 Oc­to­ber 2025) names the “autonomy in­flec­tion point” at which single-sign-on as­sump­tions break; the Model Con­text Pro­tocol's OAuth 2.1 au­thor­iz­a­tion work (ex­ten­ded Novem­ber 2025) sup­plies the mech­an­ism for is­su­ing and present­ing ma­chine cre­den­tials; the OWASP Top 10 for Agen­tic Ap­plic­a­tions (Decem­ber 2025) sup­plies the risk vocab­u­lary. What does not yet exist is gov­ernance at the scale of the pop­u­la­tion. Is­su­ing 82 cre­den­tials per human is a solved en­gin­eer­ing prob­lem; gov­ern­ing 82 cre­den­tials per human (know­ing which are act­ive, which are over-priv­ileged, which are rogue) is not.

Issuing a machine identity is solved; governing eighty-two of them per human is not.

The through-line's con­sequence for 2027 is a pro­cure­ment one. Reg­u­lated buy­ers will begin to re­quire that iden­tity ap­plies to agents as it ap­plies to people (first-class, audit­able, re­voc­able), and vendors that treat non-human iden­tity as an af­ter­thought will find it a dis­qual­i­fier. The non-human ma­jor­ity is the demo­graphic fact that makes agen­tic iden­tity a mar­ket rather than a re­search topic.

The provenance turn

The third through-line concerns content rather than identity: as generation becomes free, the scarce and valuable property is proof of origin, and provenance is turning from an opt-in feature into infrastructure.

When any image, voice or doc­u­ment can be syn­thes­ised at will, the ques­tion shifts from “is this well made” to “where did this come from.” The an­swer the in­dustry has con­verged on is cryp­to­graphic proven­ance: the Co­ali­tion for Con­tent Proven­ance and Au­then­ti­city (C2PA) and its Con­tent Cre­den­tials, which at­tach a signed, tamper-evid­ent re­cord of ori­gin and edit his­tory to a piece of con­tent. The spe­cific­a­tion reached v2.2 on 1 May 2025 and has since ad­vanced to v2.5 in Oc­to­ber 2026, with the ver­sion 2 ar­chi­tec­ture in ISO fast-track as ISO/DIS 22144; the Con­tent Au­then­ti­city Ini­ti­at­ive passed 6,000 mem­bers by Au­gust 2025; and the steer­ing com­mit­tee now in­cludes OpenAI (joined May 2024), Adobe, Google, Meta, Mi­crosoft, the BBC, Sony, Amazon and Truepic. Cru­cially, proven­ance is be­gin­ning to move from the edge into the in­fra­struc­ture: Cloud­flare began pre­serving Con­tent Cre­den­tials across its con­tent-de­liv­ery net­work in Feb­ru­ary 2025, cov­er­ing roughly a fifth of the web,

Digital TrustVolge Quarterly08 •

and cam­era makers (Leica, Nikon, Sony, Sam­sung) have begun sign­ing im­ages at the point of cap­ture.

Reg­u­la­tion is now pulling in the same dir­ec­tion. The EU AI Act's Art­icle 50 trans­par­ency duty, bind­ing from 2 Au­gust 2026 and left in­tact by the Di­gital Om­ni­bus, ob­liges dis­clos­ure of AI-gen­er­ated con­tent, with the ma­chine-read­able mark­ing duty under Art­icle 50(2) car­ry­ing a grace period to 2 Decem­ber 2026 for sys­tems already on the mar­ket: a legal de­mand that proven­ance sig­nals exist and be read­able. The com­bin­a­tion of plat­form de­fault, CDN pre­ser­va­tion, in-cam­era sign­ing and legal duty is what turns proven­ance from an opt-in badge into an in­fra­struc­tural prop­erty of the con­tent it­self.

When generation is free, the scarce asset is proof of origin.

The 2027 test, and the sub­ject of one of this edi­tion's Calls, is whether proven­ance crosses from opt-in to de­fault at scale: at least one major plat­form or CDN and one cam­era line mak­ing Con­tent Cre­den­tials the stand­ard rather than the ex­cep­tion. The turn is not yet com­plete, but its dir­ec­tion is no longer in doubt.

The post-quantum clock

The fourth through-line runs on a longer timescale but starts now: the cryptography beneath every certificate, signature and secure channel must be replaced before a quantum computer can break it, and the migration is a decade-long programme that 2027 begins in earnest.

The trust layer rests on pub­lic-key cryp­to­graphy: the math­em­at­ics be­neath PKI, di­gital sig­na­tures, TLS and the wal­lets and cre­den­tials de­scribed through­out this edi­tion. That math­em­at­ics is vul­ner­able to a suf­fi­ciently cap­able quantum com­puter, and while no such ma­chine yet ex­ists, the threat is present today through “har­vest now, de­crypt later”: an ad­versary can cap­ture en­cryp­ted traffic now and de­crypt it once the cap­ab­il­ity ar­rives. The re­sponse is a whole­sale mi­gra­tion to post-quantum al­gorithms. NIST pub­lished the first stand­ard­ised post-quantum al­gorithms (FIPS 203, 204 and 205) in Au­gust 2024, and the task since has been to move them from stand­ard into the real cer­ti­fic­ate chains, sig­na­ture schemes and pro­tocol lib­rar­ies that the trust layer de­pends on.

Volge Quarterly does not price a quantum break in­side the fore­cast win­dow; none is ex­pec­ted. The Call im­pli­cit in this through-line is subtler: that the or­gan­isa­tions build­ing trust ar­chi­tec­ture in 2027 will build it post-quantum-ready, and those bolt­ing trust on will

not, widen­ing the same ar­chi­tec­ture-versus-pos­ture gap along a di­men­sion that will not be vis­ible until it is too late to close cheaply.

This is the slow­est of the five through-lines and, for that reason, the one most eas­ily de­ferred. It is also the one with the least for­giv­ing arith­metic. A cer­ti­fic­ate is­sued in 2027 with a multi-year valid­ity, or a doc­u­ment signed today and re­lied upon for a dec­ade, must sur­vive the ar­rival of quantum cap­ab­il­ity within its own life­time. The mi­gra­tion there­fore can­not wait for the threat to be demon­strated; it must lead it. For PKI op­er­at­ors, sig­na­ture pro­viders and the sov­er­eign iden­tity schemes of India and the Gulf (all of which run na­tional pub­lic-key in­fra­struc­ture), 2027 is the year the post-quantum trans­ition moves from a re­search con­cern onto the op­er­a­tional roadmap.

Harvest now, decrypt later: the quantum threat is present today, even though the machine is not.

Assurance as return on investment

The fifth through-line reframes the economics: trust spending has been justified as loss avoidance, a cost; in 2027 it begins to be justified as a return, an enabler of transactions that could not otherwise be done.

Se­cur­ity has tra­di­tion­ally been sold on fear: the loss avoided, the breach pre­ven­ted, the fine es­caped. That fram­ing caps the budget, be­cause avoided loss is bounded by the loss it­self and is al­ways con­test­able after the fact. The ar­chi­tec­tural turn changes the eco­nom­ics. When iden­tity, re­si­li­ence and as­sur­ance are built into the product, they stop being a cost centre and be­come a cap­ab­il­ity: the abil­ity to on­board a cus­tomer in a reg­u­lated mar­ket without a branch visit, to ac­cept a European wal­let cre­den­tial the mo­ment the ac­cept­ance duty binds, to let an autonom­ous agent trans­act be­cause its iden­tity and au­thor­ity can be proved. Each of these is rev­enue that as­sur­ance en­ables, not loss it avoids.

The con­sequence for 2027 is that the trust budget es­capes its his­tor­ical ceil­ing. When as­sur­ance is a cost, it is min­im­ised; when it is a re­turn, it is in­ves­ted in. The or­gan­isa­tions that make this re­fram­ing in­tern­ally (that put the head of iden­tity in the room where growth is planned rather than the room where losses are re­viewed) will spend more on trust and get more from it. This is the through-line on which the edi­tion's op­er­ator play­book, in Part V, is built.

Digital TrustVolge Quarterly09 •

The mar­ket data sup­port the re­fram­ing. Buy­ers are real­loc­at­ing within a grow­ing en­vel­ope, Gart­ner's in­form­a­tion-se­cur­ity spend reached US$213bn in 2025 and is fore­cast near US$247.5bn for 2026 (Gart­ner, 3Q26), to­ward the iden­tity and veri­fic­a­tion lay­ers, which are grow­ing faster than the en­vel­ope it­self. That is not the be­ha­viour of a mar­ket buy­ing in­sur­ance; it is the be­ha­viour of a mar­ket buy­ing cap­ab­il­ity. The clearest ex­pres­sion is in reg­u­lated on­board­ing: iden­tity veri­fic­a­tion at US$14.34bn in 2025 grow­ing to US$29.32bn by 2030 (MarketsandMarkets, 2025) is, at root, spend on the abil­ity to ac­quire cus­tom­ers di­git­ally, not on the abil­ity to avoid fraud losses.

Assurance stops being the cost of avoiding a loss and becomes the price of doing a transaction that could not otherwise be done.
2027 is when the EU wallet stops being issued and starts being required: the year an identity credential becomes something a bank must accept, not something a customer may offer.

The buyer's ground

What the rules require, what the enforcement record costs, what is actually published about price, and how an organisation places itself against a readiness model that no supplier owns.

The calendar that binds you

Nine dated obligations fall on Gulf and Indian enterprises between February 2026 and May 2027. Six of the nine dates have already passed.

The ques­tion an en­ter­prise buyer puts to a trust out­look is not how fast the mar­ket is grow­ing. It is what the or­gan­isa­tion is re­quired to do, and by when. That ques­tion has a dated an­swer in both the Gulf and India, and the an­swer is un­usu­ally crowded. Between Feb­ru­ary 2026 and May 2027 the two re­gions im­pose bind­ing ob­lig­a­tions on au­then­tic­a­tion, fraud con­trol, sub­scriber iden­ti­fic­a­tion, syn­thetic media and per­sonal data, each rest­ing on its own in­stru­ment and each com­men­cing on its own date. The table below sets them out. It is re­stric­ted to in­stru­ments that bind. Con­sultat­ive drafts, ad­visor­ies and pub­lished roadmaps are ex­cluded, how­ever widely they have been re­por­ted.

DateInstrumentWho it bindsWhat it requires
31 Mar 2026CBUAE Notice 2025/3057UAE banks, finance companies, exchange houses, insurers, payment service providersWithdrawal of text-message and electronic-mail one-time passcodes for digital banking; liability for passcode-linked fraud rests with the institution
13 Apr 2026SAMA Counter-Fraud Fundamental RequirementsSaudi banks, finance companies, payment service providersTransaction verification beyond a passcode; re-verification on reactivation; confirmation of phone-number ownership
20 Feb 2026IT Amendment Rules, 2026 (India)Significant social media and other intermediariesLabelling and non-strippable provenance metadata on synthetically generated information; takedown within three hours
1 Apr 2026RBI authentication directionsRBI-regulated entitiesTwo distinct authentication factors for digital payment transactions unless exempted
31 Jul 2026RBI Digital Payment Security Controls DirectionsBanks, small finance banks, payments banks, urban co-operative banks, NBFCsA board-approved digital payments policy covering every digital payment product
5 Nov 2026RBI Master Direction, non-bank payment system operatorsNPCI, card networks, payment aggregators, prepaid issuersConsolidated cyber resilience and digital payment security controls
13 Nov 2026DPDP Rules, Rule 4 (India)Consent managersRegistration opens
21 Nov 2026Telecommunications (User Identification) Rules, 2026 (India)Telecom licenseesBiometric identification before a SIM is issued, changed or disconnected; the window is extendable to 21 February 2027
13 May 2027DPDP Rules 3, 5 to 16, 22 and 23 (India)All data fiduciaries, every sectorNotice, consent, security safeguards, breach notification within 72 hours, children's data, data principal rights, with penalties under sections 28 to 34

Dates shown are com­mence­ment or com­pli­ance dates as pub­lished by the is­su­ing body. Where a com­pli­ance win­dow runs from the date of no­ti­fic­a­tion, the end of that win­dow is given. The Saudi Na­tional Cy­ber­se­cur­ity Au­thor­ity's Es­sen­tial Cy­ber­se­cur­ity Con­trols, ECC-2:2024, are omit­ted from the table: the text car­ries no com­mence­ment date, no trans­ition period and no pen­alty clause of its own, re­quir­ing in­stead that com­pli­ance be con­tinu­ous. Sev­eral of the entries above rest on sec­ond­ary re­port­ing rather than on a re­trieved primary text, and are iden­ti­fied as such in the chapters that fol­low.

Three fea­tures of this cal­en­dar war­rant the at­ten­tion of a board rather than of a con­trol owner.

Digital TrustVolge Quarterly10 •
  • It is an au­then­tic­a­tion cal­en­dar be­fore it is a pri­vacy cal­en­dar. Four of the nine ob­lig­a­tions con­cern how a per­son or a device is proved, not how their data is af­ter­wards handled. The con­trol that has to change first is the login, not the con­sent no­tice.
  • It is already in force. Six of the nine dates have passed. An or­gan­isa­tion that has not yet begun is not early in a pro­gramme. It is late against a pub­lished ob­lig­a­tion.
  • Only one of the nine is a stand­ard. The re­mainder are dir­ec­tions, no­tices and rules is­sued by a reg­u­lator hold­ing su­per­vis­ory powers over the en­tit­ies they bind. A stand­ard may be ad­op­ted at a chosen pace. A dir­ec­tion may not.
Six of the nine dates have already passed. An organisation that has not begun is not early in a programme. It is late.

What the calendar does not contain

Noth­ing dated was loc­ated for either re­gion in 2028, and that ab­sence is it­self a find­ing. Post-quantum cryp­to­graphy, which oc­cu­pies a large share of vendor at­ten­tion in both mar­kets, car­ries no bind­ing Gulf or In­dian dead­line at the time of writ­ing. The United States has set 31 Decem­ber 2030 for post-quantum key es­tab­lish­ment and 31 Decem­ber 2031 for sig­na­tures at high-value fed­eral sys­tems, under the ex­ec­ut­ive order signed on 22 June 2026, leav­ing the 2035 tar­get under Na­tional Se­cur­ity Memor­andum 10 un­changed. India's Min­istry of Elec­tron­ics and In­form­a­tion Tech­no­logy and CERT-In pub­lished a post-quantum white­pa­per in July 2025 and task force re­com­mend­a­tions dur­ing 2026, both call­ing for roadmaps and cryp­to­graphic in­vent­or­ies and neither fix­ing a na­tional date. No Saudi post-quantum man­date was found. The UAE's Na­tional Post-Quantum Mi­gra­tion Pro­gramme, run by the Cyber Se­cur­ity Coun­cil with the Ad­vanced Tech­no­logy Re­search Coun­cil, is re­por­ted to re­quire gov­ern­ment en­tit­ies to file formal trans­ition plans under a Na­tional En­cryp­tion Policy tied to Fed­eral reg­u­la­tion 71 of 2024, in six man­dat­ory phases. No sub­mis­sion dead­line has been pub­licly an­nounced.

Sources for the United States ex­ec­ut­ive order dif­fer on its num­ber, some cit­ing 14412 and oth­ers 14409. The UAE en­cryp­tion policy and its phas­ing rest on sec­ond­ary re­port­ing; the primary in­stru­ment was not re­trieved.

The prac­tical con­sequence is a se­quen­cing one. Post-quantum mi­gra­tion is a genu­ine and long pro­gramme, and the or­gan­isa­tions that begin their cryp­to­graphic in­vent­ory early will fin­ish it cheaply. It is not, in either

re­gion, the ob­lig­a­tion that is cur­rently over­due. What is over­due con­cerns how cus­tom­ers and sub­scribers are au­then­tic­ated, and it fell due in the first half of 2026. A pro­gramme plan that puts cryp­to­graphic agil­ity ahead of au­then­tic­a­tion re­place­ment has read the tech­no­logy press rather than the rule­book.

Banking: the authentication deadline has passed

The UAE withdrew the one-time passcode on 31 March 2026 and placed the resulting fraud liability on the institution. Saudi Arabia and India imposed comparable obligations within a fortnight either side.

Bank­ing is the sec­tor in which the iden­tity ob­lig­a­tion has moved fur­thest from guid­ance to­ward en­force­ment, and it has done so in all three of the re­gion's prin­cipal mar­kets in­side a single quarter of 2026. The com­mon ele­ment is the re­tire­ment of the shared secret. What re­places it dif­fers by jur­is­dic­tion. That it must be re­placed does not.

United Arab Emirates

The Cent­ral Bank of the UAE re­quired that one-time passcodes de­livered by text mes­sage and elec­tronic mail be with­drawn for di­gital bank­ing by 31 March 2026, and named bio­met­rics, FIDO2 pass­keys, in-ap­plic­a­tion ap­provals and soft tokens as the ac­cept­able re­place­ments. The in­stru­ment is re­por­ted as No­tice 2025/3057, is­sued in July 2025, and its scope is re­por­ted to ex­tend bey­ond banks to fin­ance com­pan­ies, ex­change houses, in­surers and pay­ment ser­vice pro­viders. The pro­vi­sion car­ry­ing the sharpest com­mer­cial edge is not the tech­nical one. Li­ab­il­ity for fraud linked to a one-time passcode was placed on the in­sti­tu­tion. A bank that con­tin­ues to au­then­tic­ate by passcode there­fore car­ries not a su­per­vis­ory ex­pos­ure but the loss it­self.

Re­por­ted scope and the no­tice ref­er­ence vary between sec­ond­ary sources, and no primary Cent­ral Bank text was loc­ated. The ob­lig­a­tion is widely and con­sist­ently re­por­ted; the cita­tion should be con­firmed against the Cent­ral Bank's own cir­cu­lar be­fore it is re­lied on in a sub­mis­sion.

Digital TrustVolge Quarterly11 •

The po­s­i­tion of the lar­ger in­sti­tu­tions sug­gests the dead­line was met rather than con­tested. Emir­ates NBD, Abu Dhabi Is­lamic Bank and First Abu Dhabi Bank had each re­placed text-mes­sage passcodes with bio­met­ric or in-ap­plic­a­tion meth­ods for most on­line ser­vices ahead of the date, as re­por­ted in Septem­ber 2025. Sep­ar­ately, Cent­ral Bank guid­ance on di­gital iden­ti­fic­a­tion for cus­tomer due di­li­gence names three ac­cep­ted na­tional routes: the Iden­tity, Cit­izen­ship, Cus­toms and Port Se­cur­ity on­line val­id­a­tion gate­way for the Emir­ates ID, UAE PASS, and Emir­ates Fa­cial Re­cog­ni­tion. For a bank build­ing a re­mote on­board­ing jour­ney, that is a closed list rather than a mar­ket, and it is the single most con­sequen­tial fact in the chapter for any­one plan­ning a vendor se­lec­tion.

Saudi Arabia

The Saudi Cent­ral Bank is­sued Counter-Fraud Fun­da­mental Re­quire­ments on 14 April 2025, ef­fect­ive 13 April 2026, re­pla­cing the Anti-Fraud Rules of 2017 and ex­tend­ing cov­er­age to fin­ance com­pan­ies and pay­ment ser­vice pro­viders. Read op­er­a­tion­ally rather than as policy, the rule­book re­quires three things that most Gulf in­sti­tu­tions were not doing in 2024: trans­ac­tion veri­fic­a­tion must not con­sist solely of a passcode sent to the num­ber held on the ac­count; iden­tity must be re-veri­fied be­fore a ser­vice is re­act­iv­ated fol­low­ing a fraud freeze; and own­er­ship of a phone num­ber must be con­firmed as be­long­ing to the same cus­tomer. The third of these is the one that breaks ex­ist­ing ar­chi­tec­ture, be­cause it re­quires a live re­la­tion­ship with the mo­bile net­work op­er­ator rather than a field in the cus­tomer re­cord.

Ac­count­ab­il­ity is placed high. The Saudi Cyber Se­cur­ity Frame­work vests ul­ti­mate cyber re­spons­ib­il­ity with the board of the mem­ber or­gan­isa­tion and re­quires a chief in­form­a­tion se­cur­ity of­ficer at senior man­age­ment level, in­de­pend­ent of the in­form­a­tion tech­no­logy func­tion. The frame­work is re­por­ted to op­er­ate a ma­tur­ity scale from zero to five, with level three, defined as con­trols doc­u­mented, ap­proved and im­ple­men­ted, as the ex­pec­ted floor, and level four re­por­ted for se­cur­ity op­er­a­tions. Pub­lished sources dis­agree on whether levels four and five are re­quired at all, and none of this has been veri­fied against the Au­thor­ity's own text.

Liability for passcode-linked fraud now rests with the UAE institution. The control change has stopped being a compliance cost and become a loss line.

India

The Re­serve Bank has moved on three fronts in­side four­teen months. Au­then­tic­a­tion dir­ec­tions is­sued in Septem­ber 2025 re­quired at least two dis­tinct au­then­tic­a­tion factors for di­gital pay­ment trans­ac­tions from 1 April 2026, un­less ex­emp­ted. Di­gital Pay­ment Se­cur­ity Con­trols Dir­ec­tions is­sued on 31 July 2026 ex­ten­ded to com­mer­cial banks, small fin­ance banks, pay­ments banks, urban co-op­er­at­ive banks and non-bank­ing fin­an­cial com­pan­ies, cov­er­ing any di­gital pay­ment product in­clud­ing bal­ance en­quiry, set­ting and chan­ging a per­sonal iden­ti­fic­a­tion num­ber, mo­bile bank­ing re­gis­tra­tion and passcode gen­er­a­tion, and re­quir­ing a board-ap­proved di­gital pay­ments policy. A con­sol­id­ated Mas­ter Dir­ec­tion for non-bank pay­ment sys­tem op­er­at­ors fol­lowed on 5 Novem­ber 2026, bind­ing the Na­tional Pay­ments Cor­por­a­tion of India, the card net­works, pay­ment ag­greg­at­ors and pre­paid is­suers.

The con­trols named in the earlier non-bank rules in­dic­ate the dir­ec­tion of su­per­vis­ory think­ing: a cool­ing period of twelve hours after a change of re­gistered mo­bile num­ber or elec­tronic mail ad­dress, a block on re­mote-ac­cess ap­plic­a­tions dur­ing a live ses­sion, and sus­pi­cious-activ­ity alert­ing main­tained around the clock by the card net­works. Each of these is a product change rather than a se­cur­ity con­trol, which is why they ar­rive through a board-ap­proved policy rather than through a tech­nical stand­ard.

Who signs

No pub­lished source states the length of the buy­ing cycle or the sign­ing role for iden­tity pro­grammes in bank­ing in either re­gion, and none is offered here. What is pub­lished is where ac­count­ab­il­ity has been placed by in­stru­ment, and that is a bet­ter guide to the ap­proval path than a mar­ket es­tim­ate would be: the board of the mem­ber or­gan­isa­tion under the Saudi frame­work, with an in­de­pend­ent chief in­form­a­tion se­cur­ity of­ficer at senior man­age­ment level; a board-ap­proved di­gital pay­ments policy under the Re­serve Bank's dir­ec­tions of 31 July 2026. In both cases the ob­lig­a­tion has been draf­ted so that it can­not be dis­charged below board level, which means the ap­proval

Digital TrustVolge Quarterly12 •

path for the spend­ing that fol­lows runs through the same place.

The takeaway for a bank in any of the three mar­kets is that the au­then­tic­a­tion re­build is no longer a pro­ject with a busi­ness case to be made. It is an ob­lig­a­tion with a date that has passed, an owner named in the rule­book, and, in the UAE, a loss at­tached to delay.

Telecommunications: identity at the point of connection

The Gulf has made the state the verifier of record and left the operator to orchestrate around it. India has done the opposite, and charges a regulated rupee for the privilege.

Tele­com­mu­nic­a­tions is where iden­tity ob­lig­a­tions bite earli­est, be­cause the op­er­ator is the point at which a per­son first ac­quires a dur­able di­gital cre­den­tial. The two re­gions have ar­rived at op­pos­ite ar­chi­tec­tures. In the Gulf the iden­tity de­cision is made by a state plat­form and con­sumed by the op­er­ator. In India the op­er­ator per­forms the au­then­tic­a­tion it­self, against a na­tional au­thor­ity, and pays a pub­lished fee for each at­tempt.

The Gulf: the state decides

No single Gulf in­stru­ment equi­val­ent to India's sub­scriber iden­ti­fic­a­tion rules was loc­ated. What is doc­u­mented is op­er­ator prac­tice, and the prac­tice has con­verged on state iden­tity plat­forms rather than on com­mer­cial veri­fic­a­tion. Saudi Ar­a­bia re­quires a SIM to be re­gistered against a valid Iqama or bor­der num­ber with bio­met­ric veri­fic­a­tion per­formed in store. Zain's Yaqoot brand in the King­dom now re­gisters vis­itor SIMs through the Min­istry of In­terior's Ab­sher plat­form, re­mov­ing the in-branch fin­ger­print step. In Bahrain, stc im­ple­men­ted fa­cial bio­met­ric SIM ac­tiv­a­tion with the Tele­com­mu­nic­a­tions Reg­u­lat­ory Au­thor­ity and claimed a re­gional first in March 2024. In Qatar, Voda­fone Qatar de­ployed re­mote elec­tronic know-your-cus­tomer for SIM and elec­tronic SIM re­gis­tra­tion using In­nov­at­rics bio­met­rics in­teg­rated by Zentity, as re­por­ted in Decem­ber 2025.

None of the four de­ploy­ments named above car­ries a pub­lished cost, volume or out­come fig­ure. They are re­cor­ded as evid­ence of dir­ec­tion, not of re­turn. The Saudi in-store re­quire­ment rests on a con­sumer guide rather than on a re­trieved reg­u­lat­ory text.

In the Gulf the state is the verifier of record. What an operator buys is everything arranged around a decision it does not make.

The com­mer­cial con­sequence of that pat­tern is pre­cise and is routinely missed in vendor se­lec­tion. In the Gulf the op­er­ator is not per­mit­ted to be the veri­fier of re­cord. An en­ter­prise buy­ing veri­fic­a­tion tech­no­logy into Gulf tele­com­mu­nic­a­tions is there­fore buy­ing or­ches­tra­tion, live­ness, fraud scor­ing and re­cord-keep­ing ar­ranged around a na­tional de­cision, and is not buy­ing the de­cision it­self. A pro­posal priced on the as­sump­tion that the vendor per­forms the match is priced for a dif­fer­ent mar­ket.

India: the operator decides, and pays

The Tele­com­mu­nic­a­tions (User Iden­ti­fic­a­tion) Rules, 2026, no­ti­fied on 21 Au­gust 2026 by the De­part­ment of Tele­com­mu­nic­a­tions, make bio­met­ric iden­ti­fic­a­tion man­dat­ory be­fore a SIM is is­sued, changed or dis­con­nec­ted. Hold­ers of an Aadhaar num­ber are veri­fied through elec­tronic know-your-cus­tomer; a demo­graphic route ex­ists but is re­stric­ted to those without an Aadhaar num­ber or phys­ic­ally un­able to au­then­tic­ate. The com­pli­ance win­dow is three months from no­ti­fic­a­tion, ex­tend­able by up to three fur­ther months, pla­cing the op­er­at­ive date at 21 Novem­ber 2026 or, if ex­ten­ded, 21 Feb­ru­ary 2027.

A com­pan­ion cir­cu­lar of 17 Au­gust 2026 es­tab­lishes the Di­gital In­tel­li­gence Plat­form and dis­trib­utes a rep­res­ent­at­ive image of each sub­scriber to every op­er­ator daily from 24 Au­gust, en­for­cing a cap of nine con­nec­tions per sub­scriber. The ar­chi­tec­tural sig­ni­fic­ance of that cir­cu­lar ex­ceeds that of the rules them­selves. It cre­ates a shared na­tional bio­met­ric ref­er­ence set, up­dated daily, held across com­pet­ing op­er­at­ors. Whatever view is taken of the policy, an op­er­ator that has not planned for the re­ceipt, stor­age, re­ten­tion and de­le­tion of that dis­tri­bu­tion is car­ry­ing an un­quan­ti­fied ob­lig­a­tion under the data pro­tec­tion rules that com­mence on 13 May 2027.

The hole in the rules

Two gaps have been iden­ti­fied pub­licly and neither has been closed. The first is scope: the rules are si­lent on how 134.73 mil­lion ma­chine-to-ma­chine and in­ter­net-of-things con­nec­tions are to sat­isfy a bio­met­ric re­quire­ment that pre­sup­poses a human being. The second is whether call­ing ser­vices de­livered over the top of the net­work fall within the defin­i­tion at all. For an en­ter­prise run­ning a con­nec­ted fleet, a meter­ing es­tate or a lo­gist­ics op­er­a­tion, the first gap is not aca­demic. It is the dif­fer­ence between an es­tate that can be re­newed and one that can­not.

Digital TrustVolge Quarterly13 •
A biometric requirement presupposes a human being. India has 134.73 million connections that are not one, and the rules do not say what becomes of them.

The only published price in either region

In­dian tele­com­mu­nic­a­tions car­ries the only genu­inely reg­u­lated and pub­lished per-trans­ac­tion iden­tity price in the Gulf or India. Under the Aadhaar au­then­tic­a­tion fee reg­u­la­tions of 2023, the Unique Iden­ti­fic­a­tion Au­thor­ity of India charges one rupee per elec­tronic know-your-cus­tomer trans­ac­tion to tele­com ser­vice pro­viders, against three ru­pees for other re­quest­ing en­tit­ies, and fifty paise per yes-or-no au­then­tic­a­tion for au­then­tic­a­tion user agen­cies and their sub-agen­cies, in­clus­ive of taxes. A sub­sequent cir­cu­lar re­por­ted as ef­fect­ive 1 Novem­ber 2025 is said to have re­vised the elec­tronic know-your-cus­tomer fee from ₹3.40 to ₹3.60 and the yes-or-no fee to ₹0.60, char­ging for failed trans­ac­tions as well.

The re­vi­sion has not been veri­fied against the Au­thor­ity's own text and the 2023 base reg­u­la­tions should be cited in pref­er­ence until it has been.

That fig­ure mat­ters well bey­ond tele­com­mu­nic­a­tions, and it is the most use­ful num­ber in this edi­tion for any­one build­ing a busi­ness case. It is the only pub­lished unit price against which a veri­fic­a­tion volume in either re­gion can be cos­ted without a vendor quo­ta­tion, and it es­tab­lishes an order of mag­nitude against which com­mer­cial veri­fic­a­tion pri­cing, dis­cussed later in this part, sits one to two or­ders above. An en­ter­prise that can route a veri­fic­a­tion through a reg­u­lated na­tional rail rather than a com­mer­cial one is mak­ing a de­cision meas­ured in ru­pees per trans­ac­tion, not dol­lars.

Health: registration is not adoption

India has created 93.95 crore health identifiers and connected 2.72 lakh facilities. The gap between those two numbers is the whole of the opportunity and the whole of the problem.

Health­care is the sec­tor in which the re­gional iden­tity story is least about reg­u­la­tion and most about use. The in­stru­ments exist, and in India they are dated. What dis­tin­guishes health from bank­ing is that the iden­ti­fier has already been is­sued at na­tional scale and is largely not being used at the point of care. The buy­ing ques­tion in health is there­fore not how to cre­ate an iden­tity. It is how to make an iden­tity that already ex­ists ar­rive at a clin­ical work­flow.

The Mis­sion Steer­ing Group fig­ures are re­por­ted from meet­ing cov­er­age and have not been con­firmed against a gov­ern­ment re­lease.

India: the scale and the shortfall

As re­por­ted from the third Mis­sion Steer­ing Group meet­ing on 10 July 2026, 93.95 crore Ay­ush­man Bharat Health Ac­count iden­ti­fi­ers had been cre­ated and over 105 crore health re­cords linked. Against those fig­ures stand 5.33 lakh re­gistered fa­cil­it­ies, 9.85 lakh re­gistered pro­fes­sion­als, and only 2.72 lakh fa­cil­it­ies run­ning soft­ware en­abled for the Ay­ush­man Bharat Di­gital Mis­sion. A 2026 peer-re­viewed cross-sec­tional study found usage low des­pite high cre­ation, with off­line out­pa­tient re­gis­tra­tion still pre­ferred by pa­tients and staff alike.

Read as a mar­ket, the short­fall is the en­tire sub­ject. Nine hun­dred and thirty-nine mil­lion iden­ti­fi­ers have been is­sued into an es­tate where roughly half a mil­lion fa­cil­it­ies are re­gistered and barely a quarter of a mil­lion can pro­cess the iden­ti­fier in soft­ware. The con­straint is not iden­tity is­su­ance, which is solved, and it is not pa­tient will­ing­ness, which is not the bind­ing factor when the iden­ti­fier can­not be con­sumed at the desk. The con­straint is the last mile of clin­ical soft­ware.

India has issued 93.95 crore health identifiers into an estate where 2.72 lakh facilities can actually process one.

The obligation that arrives in 2027

The Di­gital Per­sonal Data Pro­tec­tion Rules, no­ti­fied on 13 Novem­ber 2025, com­mence in stages. Pro­vi­sions con­cern­ing the Data Pro­tec­tion Board took ef­fect on no­ti­fic­a­tion. Rule 4, cov­er­ing con­sent man­ager re­gis­tra­tion, com­mences on 13 Novem­ber 2026. Rules 3 and 5 to 16, to­gether with Rules 22 and 23 and the pen­alty pro­vi­sions under sec­tions 28 to 34, com­mence on 13 May 2027, bring­ing no­tice, con­sent, se­cur­ity safe­guards, breach no­ti­fic­a­tion within sev­enty-two hours, chil­dren's data and data prin­cipal rights into force for every data fi­du­ciary.

Sources dis­agree on the gaz­ette num­ber, cit­ing both G.S.R. 843(E) and G.S.R. 846(E). The gaz­ette no­ti­fic­a­tion it­self was not re­trieved and should be read be­fore the cita­tion is used.

For a hos­pital group the sig­ni­fic­ance of 13 May 2027 is spe­cific rather than gen­eral. Health data is among the most sens­it­ive cat­egor­ies a fi­du­ciary holds, chil­dren's data pro­vi­sions bear dir­ectly on pae­di­at­ric prac­tice, and a sev­enty-two hour breach no­ti­fic­a­tion ob­lig­a­tion as­sumes a de­tec­tion cap­ab­il­ity that a fa­cil­ity run­ning un­con­nec­ted soft­ware does not have. The same

Digital TrustVolge Quarterly14 •

soft­ware gap that pre­vents an iden­ti­fier being con­sumed also pre­vents a breach being seen.

The Gulf: exchange without biometrics

The Gulf has built health in­form­a­tion ex­change rather than health iden­tity. In Abu Dhabi, Malaffi matches pa­tient re­cords by Emir­ates ID rather than by bio­met­ric, and con­nects more than 3,000 fa­cil­it­ies and over 50,000 clin­ical users, hav­ing reached 3.5 bil­lion clin­ical re­cords. In Saudi Ar­a­bia, NPHIES, the na­tional health in­form­a­tion ex­change plat­form, was launched by the Coun­cil of Health In­sur­ance and the Na­tional Health In­form­a­tion Cen­ter with the Min­istry of Health on 11 March 2021. Saudi health data is per­sonal data under the Per­sonal Data Pro­tec­tion Law, and the Na­tional Data Man­age­ment Of­fice's stand­ards name both the na­tional iden­tity num­ber and health data as per­sonal data.

No pro­gramme cost is pub­lic for either Malaffi or NPHIES. No GCC health­care bio­met­ric iden­tity de­ploy­ment was found at all, des­pite tar­geted search­ing.

The Gulf ar­chi­tec­ture is the bet­ter one for a buyer to study, and for a reason that is easy to state and was hard to achieve. Match­ing on a na­tional iden­tity num­ber that every res­id­ent already holds, and that every other gov­ern­ment ser­vice already uses, re­moves the en­rol­ment prob­lem en­tirely. There is no bio­met­ric to cap­ture, no con­sent jour­ney to design for cap­ture, and no du­plic­ate-re­cord re­con­cili­ation of the kind that dom­in­ates health iden­tity pro­jects else­where. The cost of that choice is that the ex­change in­her­its whatever as­sur­ance the na­tional iden­ti­fier car­ries, and can­not ex­ceed it.

What the two regions show each other

India has solved is­su­ance and not con­sump­tion. The Gulf has solved con­sump­tion, at least in Abu Dhabi, by de­clin­ing to issue any­thing new. The les­son that travels is that a health iden­tity pro­gramme is a clin­ical soft­ware pro­gramme wear­ing an iden­tity label, and that the pub­lished counts which pro­grammes re­port, iden­ti­fi­ers cre­ated and re­cords linked, meas­ure the part of the work that is easy. The num­ber to ask for is the one India pub­lishes and most pro­grammes do not: how many fa­cil­it­ies can pro­cess the iden­ti­fier in soft­ware.

Government: sovereign identity at scale, and the machine gap

Nafath has issued 32 million digital identities and UAE PASS has over 12 million registered users. Neither scheme has published how an autonomous agent is to hold a credential.

Gov­ern­ment is the sec­tor in which the Gulf leads the re­gion out­right, and it leads on a meas­ure that mat­ters com­mer­cially: ad­op­tion. The two prin­cipal na­tional iden­tity schemes have reached a scale at which an en­ter­prise can as­sume that a Saudi or Emir­ati coun­ter­party already holds a cre­den­tial. The gap in both schemes lies else­where, and it is the gap this edi­tion's through-line is about.

Saudi Arabia

Na­fath had is­sued 32 mil­lion di­gital iden­tit­ies as stated by As­sist­ant In­terior Min­is­ter Thamer Al­harbi at a con­fer­ence in 2026. The Saudi Cent­ral Bank re­quires its su­per­vised in­sti­tu­tions to use bio­met­ric au­then­tic­a­tion at a high level of veri­fic­a­tion or above for re­mote ac­count open­ing, which places the na­tional scheme at the centre of reg­u­lated on­board­ing rather than at its peri­phery.

The Na­fath fig­ure comes from con­fer­ence re­port­ing rather than from an of­fi­cial pub­lic­a­tion. A widely cir­cu­lated claim of 4.2 mil­lion daily au­then­tic­a­tion re­quests and 800,000 at peak hour ap­pears on a single site with no of­fi­cial source and is not used here.

On the con­trol side, the Na­tional Cy­ber­se­cur­ity Au­thor­ity's Es­sen­tial Cy­ber­se­cur­ity Con­trols, ECC-2:2024, bind gov­ern­ment en­tit­ies, their con­tract­ors and op­er­at­ors of crit­ical na­tional in­fra­struc­ture. The Crit­ical Sys­tems Cy­ber­se­cur­ity Con­trols re­main at CSCC-1:2019 and have not been re­pub­lished against the 2024 re­vi­sion; com­pli­ance with the Es­sen­tial Con­trols is a pre­requis­ite for the Crit­ical Sys­tems Con­trols. Log re­ten­tion is twelve months under the former and eight­een under the lat­ter. Neither in­stru­ment car­ries a com­mence­ment date, a trans­ition period or a pen­alty clause of its own, re­quir­ing in­stead that com­pli­ance be con­tinu­ous and on­go­ing.

The ECC and CSCC de­tails above are taken from com­pli­ance guides quot­ing the pub­lished PDFs and have not been veri­fied against the Au­thor­ity's own text.

United Arab Emirates

UAE PASS re­por­ted over 12 mil­lion re­gistered users, in­clud­ing vis­it­ors, as pub­lished on 21 July 2026, al­though the Tele­com­mu­nic­a­tions and Di­gital Gov­ern­ment Reg­u­lat­ory Au­thor­ity's own counter showed a fig­ure above 11 mil­lion in Oc­to­ber 2026. The two are not re­con­ciled and both are stated here be­cause the dis­crep­ancy is in­struct­ive: a na­tional iden­tity scheme

Digital TrustVolge Quarterly15 •

that counts vis­it­ors is count­ing a dif­fer­ent pop­u­la­tion from one that counts res­id­ents, and an en­ter­prise siz­ing an ad­dress­able base should es­tab­lish which count it has been given.

The UAE has moved earli­est in the re­gion on cryp­to­graphic trans­ition. The Cyber Se­cur­ity Coun­cil partnered with the Ad­vanced Tech­no­logy Re­search Coun­cil, an­nounced 11 May 2026, and with QuantumGate on a Crypto Dis­cov­ery Tool built to Na­tional Cryp­to­graphy Cen­ter spe­cific­a­tions, map­ping and mon­it­or­ing cryp­to­graphic as­sets across crit­ical na­tional in­fra­struc­ture, an­nounced 19 May 2026. The con­tract value is not pub­lic. A Na­tional En­cryp­tion Policy is re­por­ted to re­quire gov­ern­ment en­tit­ies to file formal trans­ition plans under Fed­eral reg­u­la­tion 71 of 2024 through six man­dat­ory mi­gra­tion phases, with no sub­mis­sion dead­line pub­licly an­nounced.

India

India's gov­ern­ment in­ter­ven­tion in 2026 has been on proven­ance rather than on iden­tity. The IT Amend­ment Rules, 2026, no­ti­fied on 10 Feb­ru­ary 2026 and in force from 20 Feb­ru­ary, a win­dow of ten days, define syn­thet­ic­ally gen­er­ated in­form­a­tion as audio, image or video made or altered by com­puter so as to ap­pear real, ex­clud­ing routine edit­ing, good-faith doc­u­ments and ac­cess­ib­il­ity uses, and ex­clud­ing text gen­er­ated by ar­ti­fi­cial in­tel­li­gence. They re­quire prom­in­ent la­belling and per­man­ent proven­ance metadata that can­not be stripped, and cut the take­down win­dow for gov­ern­ment or court or­ders from thirty-six hours to three.

A fur­ther draft amend­ment was con­sul­ted on in April 2026 and an ad­vis­ory widen­ing la­belling bey­ond syn­thetic con­tent was is­sued in Oc­to­ber 2026. Neither is con­firmed as final. The gaz­ette no­ti­fic­a­tion, G.S.R. 120(E), was not re­trieved.

Both Gulf schemes are mature for citizens and silent on machines. The credential an agent holds has not been specified by either.

The gap that is common to all three

Each of these pro­grammes is ma­ture for a human being and un­defined for any­thing else. Na­fath, UAE PASS and the In­dian stack all an­swer the ques­tion of how a per­son proves who they are. None of the three has pub­lished how an autonom­ous soft­ware agent act­ing for that per­son ob­tains a cre­den­tial, how that cre­den­tial is scoped, how it is re­voked, or what re­cord is kept of what the agent did while hold­ing it. The ques­tion is not the­or­et­ical for a buyer: an en­ter­prise

that de­ploys agents against a gov­ern­ment ser­vice in either re­gion today does so by hav­ing a human cre­den­tial used by some­thing that is not a human, which is the ar­range­ment every one of these schemes was de­signed to pre­vent.

The takeaway is that the re­gional iden­tity schemes are an asset and a ceil­ing at once. They re­move the hard­est part of con­sumer on­board­ing and they do not yet reach the work­loads en­ter­prises are cur­rently build­ing. Any pro­gramme that as­sumes agent iden­tity will be sup­plied by the na­tional scheme is plan­ning against some­thing that has not been an­nounced.

The price of being late

Indian identity enforcement is measured in lakhs and UAE enforcement in hundreds of millions of dirhams. The difference is not severity of conduct. It is which obligation was breached.

Where the cost of build­ing an iden­tity pro­gramme is largely un­pub­lished, the cost of not build­ing one is a mat­ter of pub­lic re­cord. Both reg­u­lat­ors pub­lish their pen­al­ties, and the pub­lished re­cord is the most re­li­able fin­an­cial evid­ence avail­able to a buyer in either re­gion. It also con­tains a find­ing that is easy to miss and ex­pens­ive to miss: the two re­gimes pun­ish on en­tirely dif­fer­ent scales, and the scale de­pends on the in­stru­ment breached rather than on the grav­ity of the fail­ure.

The Indian record

The Re­serve Bank pub­lishes each pen­alty with the en­tity, the amount, the pro­vi­sion breached and the date. The iden­tity and know-your-cus­tomer pen­al­ties of 2026 are as fol­lows.

DateEntityAmountBreach
6 Feb 2026Bank of Maharashtra₹32.50 lakhKnow-your-customer directions
20 Mar 2026Tanur Co-operative Urban Bank₹50,000Failure to upload records to the Central KYC Records Registry
27 Mar 2026Central Bank of India₹63.60 lakhKnow-your-customer and basic savings bank deposit account directions
8 May 2026YES Bank₹31.80 lakhFailure to implement the Central KYC Records Registry identifier
Jul 2026Six non-banking financial companies₹2.70 to ₹6.20 lakh eachKnow-your-customer directions; the largest, ₹5.80 lakh, on Muthoot Finance

All of the above are taken from Re­serve Bank press re­leases. Dates for the YES Bank mat­ter are given as the date the order was re­ceived.

Digital TrustVolge Quarterly16 •

The largest iden­tity pen­alty in the In­dian re­cord for 2026 is ₹63.60 lakh, which is roughly sev­enty-two thou­sand US dol­lars. For a sched­uled com­mer­cial bank this is not a de­terrent in any fin­an­cial sense, and it would be a mis­take to treat it as one. Its func­tion is repu­ta­tional and su­per­vis­ory. The pen­alty is pub­lished with the pro­vi­sion breached, which means it be­comes part of the pub­lic su­per­vis­ory his­tory of the in­sti­tu­tion and part of the evid­ence base for the next in­spec­tion.

The Emirati record

The con­trast is stark. On 30 Oc­to­ber 2026 the Cent­ral Bank of the UAE an­nounced a pen­alty of AED 200 mil­lion on a UAE-li­censed branch of a for­eign bank for re­peated fail­ures in anti-money-laun­der­ing con­trols, gov­ernance and sanc­tions, to­gether with a man­dat­ory re­medi­ation pro­gramme under in­de­pend­ent val­id­a­tion, a pen­alty of AED 500,000 on the chief ex­ec­ut­ive and AED 300,000 on the money laun­der­ing re­port­ing of­ficer. A fur­ther pen­alty of AED 20 mil­lion is dated 24 June 2026.

Two regulators, one quarter apart. One fined a bank seventy-two thousand dollars. The other fined a bank fifty-four million, and fined two individuals as well.

Three fea­tures of the Emir­ati ac­tion change the cal­cu­la­tion for a Gulf in­sti­tu­tion. The amount is three or­ders of mag­nitude above the In­dian com­par­ator. In­di­vidu­als were pen­al­ised per­son­ally, which moves the ex­pos­ure from the bal­ance sheet to the people who sign. And the re­medi­ation pro­gramme was placed under in­de­pend­ent val­id­a­tion, which con­verts a one-off pen­alty into a multi-year, ex­tern­ally su­per­vised cost with no pub­lished ceil­ing.

Why the two records diverge

The di­ver­gence is not a state­ment about the sever­ity of the con­duct, and read­ing it that way leads to the wrong con­clu­sion. It is a state­ment about which in­stru­ment was breached. The In­dian pen­al­ties cited are for breaches of know-your-cus­tomer dir­ec­tions: a doc­u­ment­ary and re­cord-keep­ing ob­lig­a­tion. The Emir­ati pen­alty is for anti-money-laun­der­ing, gov­ernance and sanc­tions fail­ures: a pruden­tial and crim­inal-ad­ja­cent ob­lig­a­tion. The same un­der­ly­ing weak­ness, an in­sti­tu­tion that can­not re­li­ably es­tab­lish who its cus­tomer is, is pun­ished as a fil­ing fail­ure in one jur­is­dic­tion and as a con­trol fail­ure in the other.

No pub­lished fine spe­cific to iden­tity or fraud was loc­ated for Saudi Ar­a­bia. One vendor pub­lic­a­tion claims pen­al­ties from the Na­tional Cy­ber­se­cur­ity Au­thor­ity of up to SAR 25 mil­lion since Decem­ber 2024; the claim is un­cor­rob­or­ated and is not re­lied on here. Saudi Per­sonal Data Pro­tec­tion Law pen­al­ties are pub­lished as up to SAR 5 mil­lion per vi­ol­a­tion, doubled on re­pe­ti­tion.

What a board should take from the record

Two things, and the second is the one usu­ally missed. First, the pub­lished en­force­ment re­cord is the only de­fens­ible cost-of-in­ac­tion fig­ure avail­able in either re­gion, be­cause pro­gramme costs are not pub­lished and vendor es­tim­ates are not evid­ence. Second, the fig­ure that be­longs in a board paper is not the head­line pen­alty. It is the re­medi­ation pro­gramme under in­de­pend­ent val­id­a­tion, which in the Emir­ati mat­ter is open-ended and which no pub­lished source prices. An in­sti­tu­tion budget­ing for a pen­alty has budgeted for the smal­ler half of the ex­pos­ure.

What is actually published about cost

One regulated per-transaction identity price exists in the Gulf or India, and it is Indian. Everything else a buyer will be quoted is a negotiation, not a list.

This chapter states what can be sourced about the cost of di­gital trust tech­no­logy in the Gulf and India, and states plainly what can­not. The second half is longer than the first. That im­bal­ance is the find­ing, and any out­look that presents a cos­ted pro­gramme for this re­gion without say­ing so has built it from es­tim­ates.

The one regulated price

The Unique Iden­ti­fic­a­tion Au­thor­ity of India pub­lishes its au­then­tic­a­tion fees in reg­u­la­tion. Under the Aadhaar au­then­tic­a­tion fee reg­u­la­tions of 2023, elec­tronic know-your-cus­tomer au­then­tic­a­tion costs three ru­pees per trans­ac­tion for re­quest­ing en­tit­ies gen­er­ally and one rupee for tele­com ser­vice pro­viders, and yes-or-no au­then­tic­a­tion costs fifty paise, in­clus­ive of taxes. This is the only pub­lished, reg­u­lated, per-trans­ac­tion iden­tity price in either re­gion, and it is there­fore the only unit cost against which a veri­fic­a­tion volume can be mod­elled without a vendor con­ver­sa­tion.

Vendor list pricing, where it exists

Plat­form li­cens­ing is pub­lished by some vendors and not by oth­ers. Mi­crosoft pub­lishes Entra pri­cing on its own page: Entra ID P1 at six US dol­lars per user per month, P2 at nine, the Entra Suite at twelve, Entra ID Gov­ernance at seven as an add-on re­quir­ing P1 or P2, In­ter­net Ac­cess at five, Private Ac­cess at five, Per­mis­sions Man­age­ment at ten dol­lars forty cents per

Digital TrustVolge Quarterly17 •

re­source per month and Work­load ID at three dol­lars per work­load iden­tity per month. Veri­fied ID Face Check is lim­ited to eight veri­fic­a­tions per month per li­cence.

Sev­eral sec­ond­ary sources pub­lished dur­ing 2026 claim an in­crease to seven and ten dol­lars for P1 and P2 re­spect­ively from July 2026. Mi­crosoft's own pri­cing page, checked on 9 Novem­ber 2026, does not show it. The primary fig­ures are used above.

Okta's work­force iden­tity tiers are re­por­ted at six US dol­lars for Starter, four­teen for Core Es­sen­tials and sev­en­teen for Es­sen­tials per user per month billed an­nu­ally, with an an­nual con­tract min­imum of 1,500 dol­lars; Pro­fes­sional and En­ter­prise are quo­ta­tion only. These fig­ures come from third-party sum­mar­ies rather than from the vendor's own page and should be treated as in­dic­at­ive.

Per-verification pricing

Iden­tity veri­fic­a­tion is sold per veri­fic­a­tion and the pub­lished rates di­verge by more than a factor of two. Ver­iff pub­lishes 80 US cents on its Es­sen­tial tier, one dol­lar thirty-nine on Plus and one dol­lar eighty-nine on Premium, with a 49 dol­lar monthly min­imum at entry. Sum­sub pub­lishes one dol­lar thirty-five on Basic, with a 149 dol­lar monthly min­imum, and one dol­lar eighty-five on Com­pli­ance, with a 299 dol­lar min­imum. Per­sona is re­por­ted to start at about 250 dol­lars a month on a twelve-month Es­sen­tial con­tract and does not pub­lish a per-veri­fic­a­tion rate. On­fido, now part of En­trust, is quo­ta­tion only.

All four veri­fic­a­tion fig­ures come from third-party pri­cing com­par­is­ons pub­lished between July and Oc­to­ber 2026, none of which quoted the vendors' own pages, and the sources dis­agreed on Per­sona. Treat them as in­dic­at­ive rather than as list prices.

A verification costs about a dollar from a commercial provider and about a rupee from an Indian regulator. The architecture decision is worth two orders of magnitude.

Set­ting the two to­gether pro­duces the most use­ful com­par­ison in this edi­tion. A com­mer­cial veri­fic­a­tion costs on the order of one US dol­lar. A reg­u­lated In­dian elec­tronic know-your-cus­tomer au­then­tic­a­tion costs three ru­pees, and one rupee for a tele­com op­er­ator. At a volume of ten mil­lion veri­fic­a­tions a year the dif­fer­ence between rout­ing through a na­tional rail and rout­ing through a com­mer­cial pro­vider is not a pro­cure­ment sav­ing. It is an ar­chi­tec­ture de­cision that changes the order of mag­nitude of the run­ning cost.

Gart­ner's cat­egory defin­i­tions dif­fer between these up­dates and the fig­ures are not re­con­cil­able across vin­tages. One vin­tage should be cited and named.

Analyst figures

Gart­ner pro­jec­ted end-user spend­ing on in­form­a­tion se­cur­ity in the Middle East and North Africa at 4,072 mil­lion US dol­lars in 2026, a rise of 10.1 per cent on 2025, in a press re­lease dated 6 Oc­to­ber 2025. A later Gart­ner up­date puts re­gional growth at 9.1 per cent against 11.1 per cent for emer­ging Asia Pa­cific. On the global iden­tity mar­ket, Gart­ner's fourth-quarter 2025 fore­cast is re­por­ted at 18.7 bil­lion dol­lars in 2024 rising to 29.0 bil­lion by 2029, while the third-quarter 2026 fore­cast shows ac­cess man­age­ment grow­ing 9.2 per cent to 12.1 bil­lion and iden­tity gov­ernance and ad­min­is­tra­tion 10.2 per cent to 7.1 bil­lion.

What is not public, stated plainly

  • No pub­lic-sec­tor tender or con­tract award value for iden­tity, fraud or cyber work was loc­ated for Saudi Ar­a­bia, the UAE, Qatar, Bahrain, Oman, Kuwait or India. ZATCA's 2026 pro­cure­ment plan lists a tender for a sys­tem to de­tect cy­ber­at­tacks and fraud­u­lent im­per­son­a­tion of the au­thor­ity's iden­tity, with no value stated.
  • No pro­gramme cost or dur­a­tion is pub­lished for Na­fath, UAE PASS, Ab­sher, Tawakkalna, Aadhaar, the Ay­ush­man Bharat Di­gital Mis­sion or NPHIES.
  • No Gulf or In­dian bank has pub­lished an iden­tity pro­gramme budget.
  • No ana­lyst bench­mark for iden­tity pro­gramme cost or im­ple­ment­a­tion dur­a­tion spe­cific to either re­gion was loc­ated.
  • No Middle East or India iden­tity and ac­cess man­age­ment mar­ket size was loc­ated from Gart­ner or IDC; only the re­gional in­form­a­tion se­cur­ity total and the global iden­tity fig­ures above.

The con­sequence for a buyer is a pro­ced­ural one. A busi­ness case for this work in either re­gion can­not be bench­marked against pub­lished com­par­at­ors, be­cause there are none. It can be built from unit costs, which are pub­lished, from the en­force­ment re­cord, which is pub­lished, and from the or­gan­isa­tion's own volumes. A pro­posal that ar­rives with a re­gional bench­mark at­tached has sourced it from some­where that does not pub­lish, and the right ques­tion to ask is where.

Digital TrustVolge Quarterly18 •

A readiness model that is not a vendor's

Three published models exist and none belongs to a supplier. The useful one separates proofing from authentication from federation, which is the distinction most Gulf and Indian buyers conflate.

An or­gan­isa­tion that in­tends to spend on di­gital trust needs a way of stat­ing where it cur­rently stands that does not come from the party selling the rem­edy. Three such mod­els are pub­lished, two of them by na­tional bod­ies and one by a reg­u­lator, and between them they cover the ground. None of the three re­quires a li­cence and none is a ma­tur­ity as­sess­ment sold as a ser­vice.

NIST SP 800-63-4: the right spine

The Na­tional In­sti­tute of Stand­ards and Tech­no­logy pub­lished the fourth re­vi­sion of SP 800-63 in final form on 31 July 2025, with­draw­ing the third re­vi­sion on 1 Au­gust 2025. The pub­lic­a­tion com­prises a base volume to­gether with 63A-4 on en­rol­ment and iden­tity proof­ing, 63B-4 on au­then­tic­a­tion, and 63C-4 on fed­er­a­tion. Its or­gan­ising idea, and the reason it is re­com­men­ded here above the al­tern­at­ives, is that it treats three things as in­de­pend­ent that are ha­bitu­ally treated as one: the Iden­tity As­sur­ance Level, which con­cerns how well the per­son was proved at en­rol­ment; the Au­then­tic­a­tion As­sur­ance Level, which con­cerns how strongly they are proved at each sub­sequent login; and the Fed­er­a­tion As­sur­ance Level, which con­cerns how safely that proof is passed to a third party.

The fourth re­vi­sion sets each di­men­sion sep­ar­ately for each user group through a di­gital iden­tity risk man­age­ment pro­cess and re­cords the res­ult in a Di­gital Iden­tity Ac­cept­ance State­ment. It adds syn­cable au­then­tic­at­ors, which in­clude synced pass­keys, adds sub­scriber-con­trolled wal­lets to the fed­er­a­tion model, and adds con­trols for in­jec­tion at­tacks and forged media in­clud­ing deep fakes. In a fed­er­a­tion, the re­ly­ing party de­term­ines the min­imum level it will ac­cept on each of the three di­men­sions.

The level defin­i­tions sum­mar­ised here are drawn from sec­ond­ary sum­mar­ies. Be­fore any level is quoted in a design doc­u­ment the word­ing should be taken from the pub­lished volumes them­selves.

Proving who someone was at enrolment, proving they are the same person today, and passing that proof to a third party are three problems. Most programmes buy one product for all three.

The prac­tical value of the sep­ar­a­tion is im­me­di­ate in this re­gion. A Gulf bank con­sum­ing an Emir­ates ID val­id­a­tion through the na­tional gate­way has a high as­sur­ance level on proof­ing and in­her­its whatever fed­er­a­tion as­sur­ance the gate­way provides, while its own au­then­tic­a­tion as­sur­ance de­pends en­tirely on what it does at login, which until 2026 was fre­quently a text-mes­sage passcode. Stated in the model's terms, the in­sti­tu­tion was run­ning high proof­ing against low au­then­tic­a­tion, and the Cent­ral Bank's no­tice of 2026 was a cor­rec­tion to one of the three di­men­sions and not to the other two. An or­gan­isa­tion that can­not say which of its three levels is the weak one can­not se­quence its spend­ing.

CISA Zero Trust Maturity Model v2.0

The Cy­ber­se­cur­ity and In­fra­struc­ture Se­cur­ity Agency pub­lished ver­sion 2.0 in April 2023. It is or­gan­ised into five pil­lars, being Iden­tity, Devices, Net­works, Ap­plic­a­tions and Work­loads, and Data, with three cross-cut­ting cap­ab­il­it­ies: Vis­ib­il­ity and Ana­lyt­ics, Auto­ma­tion and Or­ches­tra­tion, and Gov­ernance. Each pil­lar is as­sessed across four stages: Tra­di­tional, Ini­tial, Ad­vanced and Op­timal. The agency's own an­nounce­ment re­cords that the Ini­tial stage was added in ver­sion 2.0 so that or­gan­isa­tions be­gin­ning from dif­fer­ent start­ing points could place them­selves.

On the iden­tity pil­lar, a vendor read­ing of the stages de­scribes Tra­di­tional as pass­words or basic multi-factor au­then­tic­a­tion with static rules, Ini­tial as broad multi-factor au­then­tic­a­tion with partly auto­mated en­force­ment, Ad­vanced as phish­ing-res­ist­ant meth­ods used as stand­ard, and Op­timal as con­tinu­ous re­val­id­a­tion of iden­tity within a ses­sion rather than once at sign-in. The model doc­u­ment it­self was not re­trieved and the stage word­ing should be taken from it rather than from this sum­mary.

The regulator's own tiering

The Saudi Cyber Se­cur­ity Frame­work is re­por­ted to op­er­ate a ma­tur­ity scale from zero to five, with level three, defined as con­trols doc­u­mented, ap­proved and im­ple­men­ted, as the ex­pec­ted floor for mem­ber or­gan­isa­tions, and level four re­por­ted for se­cur­ity op­er­a­tions sub­do­mains. The Counter-Fraud Frame­work is re­por­ted to use the same scale with the same ex­pect­a­tion. Pub­lished sources con­flict on whether levels four and five are re­quired at all, and none of this has been veri­fied against the Au­thor­ity's own text. The Na­tional Cy­ber­se­cur­ity Au­thor­ity's ECC-2:2024 is not tiered; CSCC-1:2019 lay­ers stricter con­trols on crit­ical sys­tems rather than de­fin­ing ma­tur­ity levels.

No iden­tity ma­tur­ity model pub­lished by Gart­ner, For­res­ter or the Cloud Se­cur­ity Al­li­ance was loc­ated. The Cloud Se­cur­ity Al­li­ance's work in this area is ratio re­search rather than a ma­tur­ity model.

Digital TrustVolge Quarterly19 •

How to use the three together

The three are not al­tern­at­ives and are best used at dif­fer­ent alti­tudes. The NIST volumes set the tar­get for a given ser­vice and user group, and are the right in­stru­ment for de­cid­ing what a par­tic­u­lar work­load re­quires. The CISA model sets the or­gan­isa­tion-wide tra­ject­ory and is the right in­stru­ment for a multi-year plan, be­cause it ac­com­mod­ates par­tial pro­gress across pil­lars. The Saudi scale, where it ap­plies, sets the su­per­vis­ory floor and is not op­tional. A buyer that has placed it­self on all three can say what a given ser­vice needs, where the or­gan­isa­tion stands over­all, and what the reg­u­lator will ac­cept, which is the whole of what a read­i­ness as­sess­ment is for.

What has actually been deployed in the region

Seven named implementations could be found across the Gulf and India. Not one of them has published a cost, a volume or a result.

A buyer asked to ap­prove an iden­tity pro­gramme will ask who else has done it. The hon­est an­swer for this re­gion is that a small num­ber of im­ple­ment­a­tions are named pub­licly, that they cluster in tele­com­mu­nic­a­tions and in state in­fra­struc­ture rather than in bank­ing, and that the pub­lished re­cord con­tains an­nounce­ments rather than out­comes. The list below is what tar­geted search­ing pro­duced. It is short, and its short­ness is it­self the find­ing.

DeploymentWhat was implementedWhenPublished outcome
Vodafone QatarRemote electronic know-your-customer for SIM and electronic SIM registration, Innovatrics biometrics integrated by ZentityDec 2025None
Zain KSA (Yaqoot)Visitor SIM registration and activation moved online through the Ministry of Interior's Absher platformNot dated in sourceNone
stc BahrainFacial biometric SIM activation with the Telecommunications Regulatory Authority, claimed a regional firstMar 2024None
Malaffi, Abu DhabiHealth information exchange matching records by Emirates IDOngoing3,000+ facilities, 50,000+ clinical users, 3.5bn clinical records; cost not public
UAE Cyber Security Council and QuantumGateNational Crypto Discovery Tool to National Cryptography Center specificationMay 2026Contract value not public
DeploymentWhat was implementedWhenPublished outcome
Indian Digital Payment Intelligence CorporationSection 8 non-profit applying AI to real-time payment fraud detection, building on the RBI's MuleHunter frameworkOct 2025Announced only; authorised capital ₹500 crore, paid-up ₹200 crore, ₹10 crore seed each from SBI and Bank of Baroda
Zain JordanIris recognition for SIM registration with IrisGuard, claimed a world first for an operatorAug 2026None; outside the Gulf and India

Every entry rests on vendor an­nounce­ments or trade re­port­ing. Where a de­ploy­ment has been an­nounced but no out­come pub­lished, that is stated rather than in­ferred.

What the list does not contain

Two ab­sences mat­ter more than the entries. Des­pite tar­geted search­ing of Emir­ates NBD, Mashreq, Riyad Bank, HDFC, ICICI and the State Bank of India, no named iden­tity veri­fic­a­tion, elec­tronic know-your-cus­tomer or deep­fake de­fence vendor con­tract could be found at any Gulf or In­dian bank. And no Gulf or In­dian hos­pital or health sys­tem bio­met­ric pa­tient iden­tity de­ploy­ment was found at all.

Bank press rooms and the Saudi Etimad pro­cure­ment portal were not reach­able through search dur­ing this re­search and should be checked dir­ectly be­fore the first ab­sence is treated as settled.

No Gulf or Indian bank has published a named identity vendor contract. The work is being done. It is not being talked about.

Reading the silence

The ab­sence of bank­ing ref­er­ences should not be read as an ab­sence of bank­ing de­ploy­ments. Three facts weigh against that read­ing. The UAE's largest in­sti­tu­tions were re­por­ted in Septem­ber 2025 to have already re­placed text-mes­sage passcodes with bio­met­ric or in-ap­plic­a­tion meth­ods. The Cent­ral Bank's dead­line of 31 March 2026 has passed without re­por­ted wide­spread non-com­pli­ance. And the Re­serve Bank's au­then­tic­a­tion dir­ec­tions took ef­fect on 1 April 2026. A great deal of au­then­tic­a­tion work has plainly been com­pleted. It has been com­pleted without pub­lic ref­er­ence cases.

Digital TrustVolge Quarterly20 •

There are or­din­ary reas­ons for that. Fraud con­trols are not ad­vert­ised, be­cause ad­vert­ising them tells the ad­versary what to work around. Vendor ref­er­ence agree­ments with reg­u­lated in­sti­tu­tions are fre­quently pro­hib­ited by the in­sti­tu­tion rather than de­clined by the vendor. And in the Gulf, where the na­tional scheme per­forms the iden­tity de­cision, there may be no vendor con­tract of the kind a ref­er­ence case de­scribes.

What a buyer should do about it

The prac­tical con­sequence is that ref­er­ence check­ing in this re­gion can­not be done through pub­lished ma­ter­ial and has to be done dir­ectly. Two ques­tions are worth more than a case study. The first is which na­tional plat­form the vendor has in­teg­rated with, by name, since Ab­sher, Na­fath, UAE PASS, the ICP gate­way and Aadhaar are dif­fer­ent in­teg­ra­tions and com­pet­ence with one does not imply com­pet­ence with an­other. The second is what the vendor's pub­lished pri­cing is, since four of the veri­fic­a­tion pro­viders named in this edi­tion pub­lish a rate and the oth­ers do not, and a vendor that will not quote a unit price in a mar­ket with no pub­lished bench­marks is ask­ing to be trus­ted on the only num­ber that can be in­de­pend­ently checked.

The machine majority, and why nobody agrees how large it is

Four vendor surveys put the ratio of machine to human identities at 109:1, 45:1, 17:1 and above 100:1. None of them has a Gulf or India cut.

The pro­pos­i­tion that ma­chine iden­tit­ies now out­num­ber human ones is the most re­peated claim in the di­gital trust mar­ket and the least evid­enced. It is re­peated be­cause it is al­most cer­tainly true in dir­ec­tion. It is un­evid­enced be­cause every fig­ure at­tached to it comes from a vendor sur­vey, the sur­veys use in­com­pat­ible defin­i­tions, and their pub­lished ra­tios dif­fer by a factor of six.

The spread is not a mat­ter of meas­ure­ment error. It is a mat­ter of defin­i­tion. A ser­vice ac­count, a con­tainer iden­tity, a work­load iden­tity, an API key, a cer­ti­fic­ate and an autonom­ous agent are all ma­chine iden­tit­ies on some defin­i­tions and not on oth­ers. An or­gan­isa­tion that counts eph­em­eral con­tainer iden­tit­ies will re­port a ratio an order of mag­nitude above one that counts only per­sist­ent ser­vice ac­counts, and both will be telling the truth about what they coun­ted.

All four are vendor sur­veys with dif­fer­ent samples and dif­fer­ent treat­ment of ar­ti­fi­cial in­tel­li­gence agents. No Gulf or India cut of any of them was found.

The four figures

SourceRatioBasis
Palo Alto Networks, 2026 Identity Security Landscape109:1Survey of 2,930 leaders; machine identities including AI agents
Cloud Security Alliance, Non-Human Identity working group45:1Median enterprise ratio, against a 17:1 baseline in 2023
Veza, State of Identity and Access 202617:1Not stated in the available summary
ManageEngineAbove 100:1 for nearly half of organisationsSome sectors reported at 500:1
The ratio is not measured. It is defined, and every vendor defines it differently.

The rule for using these numbers

If one of these fig­ures is used in a board paper, the vendor and the sample be­long in the same sen­tence as the num­ber. Writ­ten as "ma­chine iden­tit­ies out­num­ber human iden­tit­ies by 109 to one" the claim is un­us­able, be­cause it can­not be checked and does not say what was coun­ted. Writ­ten as "Palo Alto Net­works, sur­vey­ing 2,930 lead­ers in 2026, re­por­ted a ratio of 109 to one in­clud­ing AI agents" the claim is check­able, and a reader may dis­count it ap­pro­pri­ately.

What actually follows for a Gulf or Indian enterprise

The ab­sence of re­gional data does not make the prob­lem re­gional-neut­ral, and there is a spe­cific reason to ex­pect the Gulf po­s­i­tion to be un­usual. The Gulf has built its human iden­tity layer on na­tional plat­forms, which means the human side of the es­tate is com­par­at­ively well gov­erned: a res­id­ent is proved by Na­fath, by UAE PASS or by the ICP gate­way, to a stand­ard the en­ter­prise did not have to design. The ma­chine side has had no such in­ter­ven­tion. Neither Na­fath nor UAE PASS has pub­lished how a non-human actor ob­tains a cre­den­tial.

The likely con­sequence, which the pub­lished sur­veys can­not con­firm and which is offered here as an ar­gu­ment rather than as a find­ing, is that the gap between the gov­ernance of human and ma­chine iden­tity is wider in the Gulf than in mar­kets where both sides were built by the en­ter­prise. A well-gov­erned human layer and an un­gov­erned ma­chine layer is a more dan­ger­ous com­bin­a­tion than two me­diocre lay­ers,

Digital TrustVolge Quarterly21 •

be­cause the as­sur­ance of the first is routinely as­sumed to ex­tend to the second.

The practical test

The op­er­a­tional ques­tion a board can ask without any sur­vey is the one this edi­tion's through-line pro­poses: can the or­gan­isa­tion prove, at any mo­ment, who and what is act­ing in its sys­tems, and can it show the proven­ance of what those act­ors pro­duced. The second half of that ques­tion is the one that fails. An en­ter­prise that can enu­mer­ate its human iden­tit­ies to the per­son, and can­not enu­mer­ate its ser­vice ac­counts to the ac­count, has answered the ratio ques­tion for it­self and does not need a vendor sur­vey to do it.

The takeaway is there­fore to stop sourcing the ratio and start meas­ur­ing it. The num­ber that mat­ters to a board is not the in­dustry me­dian. It is the or­gan­isa­tion's own count, the pro­por­tion of that count which has an owner, and the pro­por­tion which has an ex­piry date. None of those three re­quires a sur­vey and all three can be pro­duced from ex­ist­ing dir­ect­ory and cloud tool­ing.

Six jurisdictions, one Gulf

A verification flow built around UAE PASS and the Emirates ID stalls against Nafath and the Iqama. The Gulf is one market commercially and six markets technically.

En­ter­prises that op­er­ate across the Gulf routinely plan a single iden­tity ar­chi­tec­ture for the re­gion and dis­cover, usu­ally dur­ing im­ple­ment­a­tion, that they have planned for one of six. The frag­ment­a­tion is not a mat­ter of local pref­er­ence or of vendor avail­ab­il­ity. It is struc­tural, and it runs through the iden­tity plat­form, the data trans­fer rules and the su­per­vis­ory au­thor­ity alike.

Where the fragmentation sits

Veri­fic­a­tion pro­viders op­er­at­ing across the Gulf face six sep­ar­ate reg­u­lat­ory and tech­nical en­vir­on­ments. A flow built around UAE PASS and the Emir­ates ID stalls against Na­fath and the Iqama, and Saudi data trans­fer rules are stricter than those of the UAE. The un­der­ly­ing iden­ti­fi­ers are dif­fer­ent ob­jects: the Emir­ates ID and the Iqama are is­sued by dif­fer­ent min­is­tries under dif­fer­ent legal bases, carry dif­fer­ent at­trib­utes and are val­id­ated through dif­fer­ent gate­ways, and the gate­ways do not in­ter­op­er­ate.

This char­ac­ter­isa­tion is drawn from trade re­port­ing of a vendor guide pub­lished in Au­gust 2026 and should be treated as a prac­ti­tioner's ac­count rather than as a reg­u­lat­ory state­ment.

Three layers that do not align

  • The iden­tity layer. Saudi Ar­a­bia runs Na­fath against the Iqama and the na­tional iden­tity num­ber. The UAE runs UAE PASS, the ICP val­id­a­tion gate­way and Emir­ates Fa­cial Re­cog­ni­tion, named as three dis­tinct ac­cep­ted routes in Cent­ral Bank guid­ance. Bahrain, Qatar, Oman and Kuwait each run their own. An in­teg­ra­tion is re­quired per coun­try, not per re­gion.
  • The trans­fer layer. Saudi data trans­fer rules are stricter than Emir­ati ones, which means a re­gional ar­chi­tec­ture that cent­ral­ises veri­fic­a­tion in one coun­try has to be as­sessed against the rules of the strict­est, not the most con­veni­ent.
  • The su­per­vis­ory layer. Each mar­ket has its own reg­u­lator with its own in­stru­ment and its own com­mence­ment date, as the cal­en­dar in this part sets out. A con­trol ac­cep­ted by one su­per­visor is not thereby ac­cep­ted by an­other.
The Gulf is one market when it is sold to and six when it is built for.

What this does to a programme

Three con­sequences fol­low, and all three are usu­ally dis­covered late. Costs are per coun­try rather than per re­gion, be­cause the in­teg­ra­tion work does not amort­ise. Timelines are set by the slow­est ap­proval rather than by the av­er­age, be­cause a re­gional launch gen­er­ally can­not pro­ceed while one mar­ket is un­ap­proved. And a vendor's claimed re­gional cov­er­age re­quires dis­ag­greg­a­tion, be­cause cov­er­age of the UAE is fre­quently presen­ted as cov­er­age of the Gulf, and the in­teg­ra­tion that de­liv­ers the first does not de­liver the second.

The comparison with India

India presents the op­pos­ite prob­lem and it is worth stat­ing, be­cause the con­trast cla­ri­fies what the Gulf dif­fi­culty ac­tu­ally is. India is a single jur­is­dic­tion with a single iden­tity au­thor­ity, a single pub­lished fee sched­ule and a single reg­u­lator for fin­an­cial ser­vices. The ar­chi­tec­ture is uni­form and the scale is enorm­ous. What India lacks is not con­sist­ency but head­room: 134.73 mil­lion ma­chine-to-ma­chine con­nec­tions sit out­side a bio­met­ric rule writ­ten for people, and 93.95 crore health iden­ti­fi­ers sit against 2.72 lakh fa­cil­it­ies able to pro­cess one. The In­dian prob­lem is depth of

Digital TrustVolge Quarterly22 •

ad­op­tion in­side one sys­tem. The Gulf prob­lem is breadth of in­teg­ra­tion across six.

What follows for a regional buyer

The plan­ning as­sump­tion that serves best is that the Gulf is one mar­ket com­mer­cially and six tech­nic­ally, and that the two should be budgeted sep­ar­ately. A re­gional busi­ness case built on a single in­teg­ra­tion will be wrong by roughly the num­ber of coun­tries in scope. Three ques­tions put to a vendor will es­tab­lish the po­s­i­tion quickly: which na­tional plat­forms have been in­teg­rated with, by name; which of the six mar­kets the quoted price cov­ers; and whether the pro­posed ar­chi­tec­ture cent­ral­ises veri­fic­a­tion in a jur­is­dic­tion whose trans­fer rules are more per­missive than those of the coun­try whose res­id­ents are being veri­fied.

None of this ar­gues against a re­gional ar­chi­tec­ture. It ar­gues for one that is de­signed from the out­set as six in­teg­ra­tions be­hind a com­mon or­ches­tra­tion layer, which is a dif­fer­ent and more ex­pens­ive thing than a single in­teg­ra­tion as­sumed to travel.

The Gulf, country by country

Digital trust in each of the six Gulf states: the scheme, the instrument that binds, what has been enforced, and what no regulator in the region has yet published.

Saudi Arabia: four instruments and one unreliable number

Saudi Arabia is the only Gulf state in which identity, fraud, cryptography and data enforcement each rest on a dated and retrievable instrument. The weakest element of the position is the one most often quoted in its favour: the user count.

The po­s­i­tion in Saudi Ar­a­bia is that the reg­u­lat­ory ap­par­atus around di­gital trust is now the most com­plete in the Gulf, and that the fig­ures de­scrib­ing its reach are the least re­li­able part of it. Four in­stru­ments bind or pur­port to bind, each car­ry­ing its own date: a Counter-Fraud Frame­work of Oc­to­ber 2022, Counter-Fraud Fun­da­mental Re­quire­ments of April 2025, a quantum re­si­li­ence cir­cu­lar of Au­gust 2026, and deep­fake guidelines of May 2026. Against that, the num­ber most often cited in sup­port of the King­dom's iden­tity in­fra­struc­ture, the Na­fath user count, is re­por­ted in at least two in­com­pat­ible forms and without any defin­i­tion of what is being coun­ted. It is to be noted that a buyer should plan on the in­stru­ments, which are dated, and not on the count, which is not.

The scheme it­self is Na­fath, also de­scribed as Uni­fied Na­tional Ac­cess. It is op­er­ated by the Na­tional In­form­a­tion Cen­ter, with the Saudi Data and Ar­ti­fi­cial In­tel­li­gence Au­thor­ity ex­er­cising over­sight of the na­tional di­gital plat­forms. It serves as the login route into gov­ern­ment and, in­creas­ingly, private sec­tor ser­vices, and as the au­then­tic­a­tion factor for trans­ac­tions re­quir­ing a second step. On is­sued iden­tit­ies, the As­sist­ant Min­is­ter of In­terior for Tech­no­logy Af­fairs stated in 2026 that 32 mil­lion di­gital iden­tit­ies had been is­sued through the plat­form. Min­istry state­ments re­por­ted at LEAP 2026 put the fig­ure at 32 mil­lion be­ne­fi­ciar­ies, linked to 1,245 plat­forms and sup­port­ing more than 6.5 bil­lion trans­ac­tions. An earlier fig­ure at­trib­uted to the Au­thor­ity, dated Novem­ber 2024, gives more than 23.5 mil­lion users and more than 3 bil­lion veri­fic­a­tion op­er­a­tions.

The Na­fath fig­ures above come from re­port­ing of min­istry state­ments at LEAP 2026 and at the Se­cur­ity and His­tory Dia­logue 2026, and from a third-party guide cit­ing the Au­thor­ity for the Novem­ber 2024 po­s­i­tion. Di­gital iden­tit­ies is­sued, be­ne­fi­ciar­ies and users are three dif­fer­ent met­rics and no pub­lished defin­i­tion re­con­ciles them. They are un­re­con­ciled, and no pen­et­ra­tion rate has been built on them.

What binds, and from when

The au­thor­ity that reg­u­lates iden­tity, au­then­tic­a­tion and fraud for fin­an­cial ser­vices is the Saudi Cent­ral Bank. The bind­ing in­stru­ment on fraud is the Counter-Fraud Frame­work, ver­sion 1.0, dated Oc­to­ber 2022 and pub­lished in the Bank's rule­book. Its cir­cu­lar set a se­quence rather than a single date: gap as­sess­ment with monthly re­port­ing from 30 Novem­ber 2022, a roadmap to at least the third ma­tur­ity level by the same date, full com­pli­ance by 29 June 2023, and an an­nual in­ternal audit re­port from the end of the fourth quarter of 2023. It op­er­ates on a ma­tur­ity model run­ning from zero to five and re­quires a ded­ic­ated counter-fraud func­tion, an en­ter­prise-wide fraud risk as­sess­ment and im­me­di­ate no­ti­fic­a­tion of new ty­po­lo­gies. The Counter-Fraud Fun­da­mental Re­quire­ments that fol­lowed on 14 April 2025, re­por­ted ef­fect­ive 13 April 2026, ex­ten­ded the re­gime to the wider fin­an­cial sec­tor and ob­lige in­sti­tu­tions to re­port the value of cus­tomer and op­er­a­tional fraud losses. No ag­greg­ate from that re­port­ing has been pub­lished, so the King­dom now col­lects the fig­ures that would settle the ques­tion of Gulf fraud losses and has not re­leased them.

  • Counter-Fraud Frame­work ver­sion 1.0, Saudi Cent­ral Bank, Oc­to­ber 2022. Full com­pli­ance due 29 June 2023.
Digital TrustVolge Quarterly23 •
  • Counter-Fraud Fun­da­mental Re­quire­ments, is­sued 14 April 2025, re­por­ted ef­fect­ive 13 April 2026. Loss val­ues re­port­able to the Bank.
  • En­hance­ment of Op­er­a­tional Re­si­li­ence to Ad­dress Quantum Com­put­ing Risks, Saudi Cent­ral Bank, 27 Au­gust 2026. Cryp­to­graphic asset iden­ti­fic­a­tion and clas­si­fic­a­tion by end of Q4 2026; en­ter­prise quantum risk as­sess­ment by end of Q1 2027.
  • Deep­fakes Guidelines, Saudi Data and Ar­ti­fi­cial In­tel­li­gence Au­thor­ity, 8 May 2026. Wa­ter­mark­ing, dis­clos­ure and con­sent; use for fraud, im­per­son­a­tion and de­fam­a­tion pro­hib­ited.
  • Per­sonal Data Pro­tec­tion Law, Art­icle 36. Forty-eight vi­ol­a­tion de­cisions an­nounced 16 Janu­ary 2026.

On re­mote on­board­ing, the pub­lished po­s­i­tion is thin­ner than the fraud po­s­i­tion. A Cent­ral Bank cir­cu­lar on bio­met­ric au­then­tic­a­tion for re­mote cus­tomer re­la­tion­ship ini­ti­ation is re­ferred to in com­ment­ary, and re­mote on­board­ing re­quire­ments are also de­scribed as sit­ting within the Bank's anti-money-laun­der­ing rules. The fre­quently re­peated ac­count that fully di­gital on­board­ing is per­mit­ted for lower ac­count tiers where iden­tity is checked through Na­fath for na­tion­als and Ab­sher for res­id­ents could not be traced to an of­fi­cial text. What can be stated is that the iden­tity route into Saudi on­board­ing is a gov­ern­ment route, that the cre­den­tial is the na­tional iden­tity num­ber or the Iqama re­cord, and that veri­fic­a­tion is a real-time match against the na­tional re­gister rather than a doc­u­ment scan.

The ac­count of the Saudi elec­tronic know-your-cus­tomer route in the pre­ced­ing para­graph rests on vendor and con­sultancy com­ment­ary, not on a re­trieved Cent­ral Bank cir­cu­lar. It should not be treated as es­tab­lished. The fraud and quantum in­stru­ments, by con­trast, are pub­lished in the Bank's own rule­book.

Cryptography, synthetic media and enforcement

The quantum po­s­i­tion is the most con­crete in the re­gion. The Cent­ral Bank's cir­cu­lar En­hance­ment of Op­er­a­tional Re­si­li­ence to Ad­dress Quantum Com­put­ing Risks, dated 27 Au­gust 2026, re­quires that pro­ced­ures for identi­fy­ing and clas­si­fy­ing all cryp­to­graphic as­sets be ac­cur­ate and com­pre­hens­ive by the end of the fourth quarter of 2026, that as­so­ci­ated data, sys­tems and ser­vices be clas­si­fied by sens­it­iv­ity and mi­gra­tion pri­or­ity, and that an en­ter­prise-level quantum risk as­sess­ment with ac­tion plans be com­pleted by the end of the first quarter of 2027. It sets no mi­gra­tion end-date and re­quires no sub­mis­sion: ex­am­iners may re­quest the in­vent­ory and the as­sess­ment. The Na­tional Cryp­to­graphic Stand­ards is­sued by the Na­tional Cy­ber­se­cur­ity Au­thor­ity are sep­ar­ately re­por­ted to

re­quire that cryp­to­graphic sys­tems sup­port post-quantum al­gorithms, in the present tense and without a phase-in.

The Kingdom now collects the fraud loss figures that would settle the Gulf argument, and has not published them.

On syn­thetic media, the Au­thor­ity is­sued Deep­fakes Guidelines on 8 May 2026. The doc­u­ment defines deep­fakes as con­tent ma­nip­u­lated through deep learn­ing, calls for vis­ible and tamper-res­ist­ant wa­ter­mark­ing, re­quires cre­at­ors to dis­close deep­fake use in de­scrip­tions or cred­its, re­quires ex­pli­cit con­sent and the pro­tec­tion of per­sonal data, pro­hib­its use for fraud, im­per­son­a­tion or de­fam­a­tion, and points read­ers to­wards de­tec­tion and proven­ance tool­ing. Whether the guidelines bind is con­tested: the state news agency de­scribes the doc­u­ment as man­dat­ing ad­her­ence, while at least one reg­u­lat­ory tracker clas­si­fies it as non-bind­ing guid­ance in a jur­is­dic­tion with no AI stat­ute. Both read­ings are un­re­con­ciled.

En­force­ment is pub­lished, in part. On 16 Janu­ary 2026 the state news agency re­cor­ded that the Au­thor­ity's com­mit­tees for re­view­ing vi­ol­a­tions had is­sued forty-eight de­cisions con­firm­ing breaches and im­pos­ing the pen­al­ties pre­scribed under Art­icle 36 of the data pro­tec­tion law. The vi­ol­a­tion types were named: col­lect­ing and pro­cessing per­sonal data, dis­clos­ing it without legal jus­ti­fic­a­tion, fail­ure to im­ple­ment ap­pro­pri­ate or­gan­isa­tional, ad­min­is­trat­ive and tech­nical meas­ures, and send­ing mar­ket­ing mes­sages without con­sent. The amounts were not dis­closed, and the ceil­ings given in sec­ond­ary sources are un­re­con­ciled.

On ma­chine or agent iden­tity, noth­ing has been pub­lished by the Cent­ral Bank, the Au­thor­ity or the Cy­ber­se­cur­ity Au­thor­ity. That is the find­ing, and it holds across all six states.

What the Saudi po­s­i­tion asks of a buyer is a re­order­ing of ef­fort. The dead­lines that fall first are cryp­to­graphic, not bio­met­ric: a cryp­to­graphic asset in­vent­ory is due within weeks of this edi­tion, and an en­ter­prise quantum risk as­sess­ment within months. The fraud ob­lig­a­tions are already in force and already re­quire loss re­port­ing, which means the in­ternal num­ber ex­ists even where the na­tional num­ber does not. And the iden­tity reach of Na­fath, whatever its true mag­nitude, should be treated in plan­ning as a route rather than as a pop­u­la­tion.

The scheme is UAE PASS, the na­tional di­gital iden­tity, ad­min­istered in con­junc­tion with the

Digital TrustVolge Quarterly24 •

Tele­com­mu­nic­a­tions and Di­gital Gov­ern­ment Reg­u­lat­ory Au­thor­ity and used across fed­eral and emir­ate-level ser­vices. Its re­gistered user fig­ure is re­por­ted in three in­com­pat­ible forms within a few months. The Au­thor­ity's own page is de­scribed as list­ing more than el­even mil­lion re­gistered users, more than fif­teen thou­sand ser­vices, 363 ser­vice pro­viders and 1,028 di­gital chan­nels. A World Eco­nomic Forum re­cog­ni­tion piece gives more than twelve mil­lion. A Di­gital Dubai so­cial media post in late July 2026 claims more than four­teen mil­lion.

The United Arab Emirates: the authentication rule that has already passed

The Emirates is the only Gulf state to have withdrawn a mainstream authentication method by regulatory deadline, and that deadline fell in March 2026. The published record of what followed is almost entirely absent.

The po­s­i­tion in the United Arab Emir­ates is that the most con­sequen­tial di­gital trust in­stru­ment in the Gulf has already taken ef­fect, and that its out­come can­not be veri­fied. The Cent­ral Bank re­quired li­censed fin­an­cial in­sti­tu­tions to with­draw one-time passcodes de­livered by text mes­sage and elec­tronic mail for di­gital bank­ing by 31 March 2026, and to re­place them with bio­met­ric and in-ap­plic­a­tion meth­ods. That date has passed. No post-dead­line com­pli­ance state­ment, en­force­ment no­tice or ex­emp­tion re­gister has been loc­ated. It is to be noted that an en­ter­prise which as­sumes the rule was en­forced is as­sum­ing it on the bal­ance of re­port­ing, not on the re­cord.

The el­even, twelve and four­teen mil­lion fig­ures are un­re­con­ciled. The highest of them rests on a so­cial media post. The re­la­tion­ship between re­gistered ac­counts and dis­tinct res­id­ent per­sons is not pub­lished, and no pen­et­ra­tion rate should be de­rived from any of the three.

What UAE PASS is ac­cep­ted for runs wider than gov­ern­ment. The Cent­ral Bank has per­mit­ted li­censed fin­an­cial in­sti­tu­tions to use it to verify cus­tomer iden­tity, and its sig­na­ture func­tion car­ries legal weight for elec­tronic ex­e­cu­tion. The bio­met­ric step is a face or fin­ger­print check against the Emir­ates iden­tity re­cord held in gov­ern­ment sys­tems rather than against a doc­u­ment presen­ted by the cus­tomer.

The instrument, and what is missing around it

The reg­u­lat­ing au­thor­ity for iden­tity, au­then­tic­a­tion and fraud in fin­an­cial ser­vices is the Cent­ral Bank of the United Arab Emir­ates. The bind­ing in­stru­ment on au­then­tic­a­tion is re­por­ted as No­tice 2025/3057, cov­er­ing banks, fin­ance com­pan­ies, ex­change houses, in­surers and pay­ment ser­vice pro­viders. Its sub­stance is the with­drawal of text-mes­sage and elec­tronic-mail

passcodes for di­gital bank­ing by 31 March 2026, their re­place­ment by risk-based meth­ods in­clud­ing bio­met­ric checks, hard­ware and soft­ware tokens, in-ap­plic­a­tion ap­proval and pass­keys, a com­pli­ance re­view at six-monthly in­ter­vals, and a shift of li­ab­il­ity for passcode-linked fraud onto the in­sti­tu­tion. Re­port­ing gives a ceil­ing of 250,000 dirhams for sig­ni­fic­ant breaches.

Each of those par­tic­u­lars rests on sec­ond­ary re­port­ing. The issue date is given vari­ously as May 2025, June 2025 and 25 July 2025, and the no­tice it­self was not re­trieved. Major banks are re­por­ted to have dis­abled text-mes­sage passcodes for on­line card pay­ments and di­gital bank­ing from 6 Janu­ary 2026, ahead of the dead­line. A second no­tice, numbered 2176 and dated April 2026, is de­scribed as im­pos­ing some twenty re­quire­ments con­cern­ing mal­ware on the An­droid plat­form; that de­scrip­tion is a vendor sum­mary.

No­tice 2025/3057 and No­tice 2176/2026 are de­scribed here from sec­ond­ary re­port­ing and vendor com­ment­ary. Neither no­tice was re­trieved from the Cent­ral Bank. The con­flict­ing issue dates for the first are re­pro­duced rather than re­solved.

On elec­tronic know-your-cus­tomer the Emir­ates has the clearest primary doc­u­ment in the Gulf. The Cent­ral Bank's Guid­ance for Li­censed Fin­an­cial In­sti­tu­tions on Di­gital Iden­ti­fic­a­tion for Cus­tomer Due Di­li­gence, is­sued by the Anti-Money Laun­der­ing and Com­bat­ing the Fin­an­cing of Ter­ror­ism Su­per­vi­sion De­part­ment and dated 31 Oc­to­ber 2022, cov­ers iden­tity proof­ing and en­rol­ment, au­then­tic­a­tion and iden­tity li­fe­cycle man­age­ment, and port­ab­il­ity and in­ter­op­er­ab­il­ity. It names UAE PASS and the fed­eral iden­tity au­thor­ity's on­line val­id­a­tion gate­way among the ac­cep­ted routes, and re­quires in­sti­tu­tions to re­tain the Emir­ates iden­tity doc­u­ment and the re­cord of its di­gital veri­fic­a­tion. The route is per­mit­ted rather than man­dated, so the li­cence, the risk pro­file and su­per­vis­ory ex­pect­a­tion de­term­ine how far a fully re­mote flow may go.

The Emirates withdrew an authentication method on a published date and published nothing about what happened next.

Cryptography, synthetic media, penalties

The cryp­to­graphic pro­gramme is na­tional, named and pro­cured. A Na­tional En­cryp­tion Policy ap­proved in 2025 re­quires gov­ern­ment en­tit­ies to hold form­ally ap­proved plans for mi­gra­tion to post-quantum cryp­to­graphy, with cryp­to­graphic dis­cov­ery and in­vent­ory as the pre­con­di­tion. On 11 May 2026 the Cyber Se­cur­ity Coun­cil signed an agree­ment with the com­mer­cial arm

Digital TrustVolge Quarterly25 •

of the Ad­vanced Tech­no­logy Re­search Coun­cil to de­ploy and scale named com­pon­ents: a crypto dis­cov­ery tool cus­tom­ised to re­quire­ments set by the Na­tional Cryp­to­graphy Cen­ter, en­tan­gle­ment-based quantum key dis­tri­bu­tion already used to con­nect data centres at one Abu Dhabi en­tity, and na­tional cryp­to­graphic lib­rar­ies con­tain­ing both clas­sical and post-quantum al­gorithms. No sub­mis­sion dead­line for the mi­gra­tion plans has been pub­lished, and the two ag­greg­at­ors that re­cord the po­s­i­tion dis­agree on whether one ex­ists.

On syn­thetic media the po­s­i­tion is ad­vis­ory and dated. The Na­tional Pro­gramme for Ar­ti­fi­cial In­tel­li­gence pub­lished a Deep­fake Guide in July 2021 which re­com­mends wa­ter­marks, cryp­to­graphic sig­na­tures and metadata to es­tab­lish con­tent au­then­ti­city and label syn­thetic ma­ter­ial. It is not bind­ing law. As of Au­gust 2026 no stan­dalone deep­fake stat­ute had been en­acted, and cre­at­ors of syn­thetic media are not re­quired to label it as ar­ti­fi­cially gen­er­ated. Con­duct is pro­sec­uted under the cy­ber­crime de­cree-law of 2021, under which de­fam­a­tion using ma­nip­u­lated con­tent is re­por­ted to carry im­pris­on­ment of at least one year and a fine of between 250,000 and 500,000 dirhams. Re­search cur­rent to March 2026 found no re­por­ted ju­di­cial de­cision on the ad­miss­ib­il­ity of a deep­fake as evid­ence.

Pen­al­ties, by con­trast, are pub­lished and large. On 30 Oc­to­ber 2026 the Cent­ral Bank im­posed a pen­alty of 200 mil­lion dirhams on the branch of a for­eign bank for fail­ures across anti-money-laun­der­ing, gov­ernance, com­pli­ance and sanc­tions con­trols, with 500,000 dirhams on the chief ex­ec­ut­ive and 300,000 dirhams on the head of com­pli­ance, and an order re­quir­ing a man­dat­ory re­medi­ation pro­gramme. On 24 June 2026 a pen­alty of 20 mil­lion dirhams was im­posed on an­other for­eign bank branch. A re­view of the Cent­ral Bank's own re­leases puts pen­al­ties on in­sti­tu­tions and of­fi­cials at ap­prox­im­ately 222.92 mil­lion dirhams between Janu­ary and Oc­to­ber 2026, against 214.5 mil­lion dirhams for 2025. The in­sti­tu­tions were not named.

On ma­chine or agent iden­tity noth­ing has been pub­lished. Neither the Cent­ral Bank nor the Cyber Se­cur­ity Coun­cil has is­sued an in­stru­ment, guideline or con­sulta­tion ad­dress­ing the iden­tity of an autonom­ous soft­ware agent.

What the Emir­ates po­s­i­tion asks of a buyer is evid­en­tiary dis­cip­line rather than new tech­no­logy. The au­then­tic­a­tion mi­gra­tion is be­hind, not ahead, and the ques­tion now is whether the in­sti­tu­tion can pro­duce, on re­quest, the re­cord of what it with­drew, what it sub­sti­tuted and when. The pen­alty re­cord shows a reg­u­lator that fines in­di­vidu­als as well as in­sti­tu­tions and pub­lishes the amounts without nam­ing the firm, which cal­ib­rates the de­terrent to the of­ficer rather than to the brand. And con­tent proven­ance re­mains the weak­est limb: the re­com­men­ded con­trols are five years old and ad­vis­ory, and no la­belling duty ex­ists to lean on.

Qatar and Bahrain: one rule each, and silence around it

Qatar and Bahrain each have a dated financial-sector onboarding rule that can be traced to the regulator, and almost nothing else. Neither publishes identity adoption figures, fraud losses, provenance requirements or a cryptographic programme.

The po­s­i­tion in Qatar and Bahrain is that each has done one thing well and pub­lished noth­ing around it. Qatar's cent­ral bank is­sued elec­tronic know-your-cus­tomer pro­ced­ures on 15 Oc­to­ber 2023. Bahrain's cent­ral bank man­dated a na­tional elec­tronic know-your-cus­tomer plat­form, op­er­ated by the na­tional pay­ments com­pany with the elec­tronic gov­ern­ment au­thor­ity, live with its first bank on 30 April 2019 and writ­ten into the cent­ral bank's fin­an­cial crime rule­book with named ex­cep­tions. Bey­ond those two in­stru­ments the re­cord thins to noth­ing: no pub­lished user count for either na­tional iden­tity scheme, no fraud loss fig­ure, no deep­fake or proven­ance meas­ure, no cryp­to­graphic pro­gramme, and no en­force­ment ac­tion touch­ing iden­tity or fraud. It is to be noted that this is a find­ing about dis­clos­ure, not an in­fer­ence about cap­ab­il­ity.

Qatar

Qatar op­er­ates two iden­tity as­sets whose re­la­tion­ship has not been of­fi­cially stated. The first is Tawtheeq, whose Eng­lish name is the Na­tional Au­then­tic­a­tion Sys­tem, launched by the min­istry re­spons­ible for trans­port and com­mu­nic­a­tions and dated by sec­ond­ary guides to 2019. Its func­tion is login and iden­tity man­age­ment for gov­ern­ment elec­tronic ser­vices, sup­port­ing a user­name and pass­word with a second factor, or a smart iden­tity card. The second is the Qatar Di­gital ID ap­plic­a­tion, at­trib­uted to the Min­istry of In­terior and de­scribed as a cre­den­tial wal­let, re­por­ted as en­abled for air­port elec­tronic gate use by early 2025. No of­fi­cial state­ment that the second re­places the first

Digital TrustVolge Quarterly26 •

has been loc­ated, and no re­gistered user fig­ure has been pub­lished for either.

The au­thor­ity that reg­u­lates iden­tity and au­then­tic­a­tion for fin­an­cial ser­vices is the Qatar Cent­ral Bank, which su­per­vises banks, Is­lamic banks, in­surers and ex­change com­pan­ies, with en­tit­ies li­censed in­side the Qatar Fin­an­cial Centre fall­ing to that centre's own reg­u­lator. The bind­ing in­stru­ment is the elec­tronic know-your-cus­tomer pro­ced­ures, which the state news agency re­cor­ded as is­sued on Sunday 15 Oc­to­ber 2023, with the stated pur­pose of reg­u­lat­ing and im­prov­ing di­gital veri­fic­a­tion of cus­tomer iden­tity and en­sur­ing full com­pli­ance with the rules against money laun­der­ing and ter­ror­ist fin­an­cing. One com­pli­ance com­ment­ary gives 9 Oc­to­ber 2023 as the date of the reg­u­la­tion it­self, with the an­nounce­ment fol­low­ing on the fif­teenth; the two dates are un­re­con­ciled.

What the ac­cep­ted routes are can be es­tab­lished only in part. The bank's Pay­ment Ser­vices Reg­u­la­tion re­quires li­censees to com­ply with the se­cur­ity prin­ciples for elec­tronic know-your-cus­tomer when on­board­ing di­git­ally, and lists the ac­cep­ted doc­u­ments: a Min­istry of In­terior iden­tity card for cit­izens and res­id­ents, a com­mer­cial re­gis­tra­tion cer­ti­fic­ate for com­pan­ies, and a pass­port for vis­it­ors and non-res­id­ents. No pub­lished text ties either Tawtheeq or the Qatar Di­gital ID to the cent­ral bank's on­board­ing route, which is a sig­ni­fic­ant gap: the cre­den­tial and the rule exist, and the bridge between them is not doc­u­mented in Eng­lish.

The live­ness re­quire­ment at­trib­uted to the Qatari pro­ced­ures, and the five per cent ad­op­tion fig­ure for the Qatar Di­gital ID, both rest on single vendor sources and are re­por­ted here as un­veri­fied. The issue date of 15 Oc­to­ber 2023 is from the state news agency and is primary.

On syn­thetic media Qatar has warn­ings rather than in­stru­ments. In Au­gust 2026 the Min­istry of In­terior cau­tioned that deep­fake and ar­ti­fi­cially gen­er­ated con­tent was being used to im­per­son­ate in­di­vidu­als, and that con­tent should not be trus­ted or shared until its source and au­then­ti­city have been veri­fied. Re­port­ing of the ad­vis­ory noted that shar­ing mis­lead­ing in­form­a­tion, de­lib­er­ately or neg­li­gently, could carry con­sequences under the ex­ist­ing cy­ber­crime law. The Na­tional Cyber Se­cur­ity Agency is­sued sep­ar­ate 2026 alerts on im­per­son­a­tion of trus­ted gov­ern­ment en­tit­ies and major com­pan­ies, and on fraud­u­lent web­sites. No proven­ance, wa­ter­mark­ing or la­belling re­quire­ment, no cryp­to­graphic or post-quantum pro­gramme, no en­force­ment ac­tion touch­ing iden­tity or fraud, and noth­ing on ma­chine or agent iden­tity has been pub­lished.

The au­thor­ity that binds is the Cent­ral Bank of Bahrain, and here the in­stru­ment is un­usu­ally spe­cific. The fin­an­cial crime mod­ule of the rule­book, in both the con­ven­tional and Is­lamic bank­ing volumes, pro­hib­its

li­censees from open­ing ac­counts for nat­ural per­sons resid­ing out­side the Gulf Co­oper­a­tion Coun­cil through di­gital on­board­ing, with two ex­cep­tions: hold­ers of the Golden Visa, and na­tion­als of Coun­cil states hold­ing a Bahraini per­sonal iden­tity num­ber who use the Na­tional eKYC sys­tem. Where a li­censee col­lects a sig­na­ture elec­tron­ic­ally it must com­ply with the Elec­tronic Trans­ac­tions Law of 2018. Where it op­er­ates its own pro­pri­et­ary di­gital iden­tity ap­plic­a­tion it must sub­mit that ap­plic­a­tion for re­view, and the ap­plic­a­tion must carry geo­loca­tion fea­tures so that the on­board­ing loc­a­tion can be checked. The mod­ule re­cords amend­ments in 2022 and 2023, the cap­ital mar­kets volume added di­gital on­board­ing pro­vi­sions in Oc­to­ber 2023, and a let­ter on the na­tional plat­form ap­pears in the bank's index under 17 Feb­ru­ary 2025.

Bahrain

Bahrain's scheme is the eKey, form­ally the Na­tional Au­then­tic­a­tion Frame­work, is­sued by the In­form­a­tion and eGovernment Au­thor­ity. It func­tions as a uni­fied login and as a na­tional di­gital iden­tity which au­then­tic­ates user data. It is tiered: a stand­ard ac­count re­quires only a per­sonal num­ber and pass­word and can be cre­ated on­line, while an ad­vanced ac­count re­quires an iden­tity card and fin­ger­print veri­fic­a­tion at a kiosk or ser­vice centre and un­locks the full range of ser­vices. The second ver­sion is a single sign-on plat­form whose first phase covered ap­prox­im­ately four­teen gov­ern­ment and private sec­tor en­tit­ies. A busi­ness-fa­cing vari­ant provid­ing bio­met­ric iden­tity and know-your-cus­tomer ser­vices for firms is de­livered through a com­mer­cial sub­si­di­ary, and the Na­tional Bank of Bahrain de­ployed it for pass­word­less di­gital iden­tity in Decem­ber. No re­gistered user count has been pub­lished.

Bahrain's onboarding rule names the exceptions and the conditions; what it does not do is publish a number.

What the two po­s­i­tions ask of a buyer is a change in the kind of evid­ence sought. In Qatar and Bahrain the reg­u­lat­ory ques­tion that can be answered is nar­row and doc­u­ment­ary: which on­board­ing routes are per­mit­ted, which res­id­ency cat­egor­ies are ex­cluded, and what con­di­tions at­tach to a pro­pri­et­ary iden­tity ap­plic­a­tion. The ques­tions that can­not be answered from pub­lished sources are the ones a risk com­mit­tee asks first: how many people hold the cre­den­tial, how much is being lost to fraud, and what hap­pens when a

Digital TrustVolge Quarterly27 •

con­trol fails. Those should be put to the coun­ter­party and the reg­u­lator dir­ectly, and the ab­sence of an an­swer re­cor­ded as an ab­sence rather than filled by a re­gional av­er­age.

The plat­form be­hind that rule is op­er­ated by the na­tional pay­ments com­pany with the elec­tronic gov­ern­ment au­thor­ity under cent­ral bank su­per­vi­sion, and gives fin­an­cial in­sti­tu­tions a na­tional iden­tity data­base for veri­fy­ing cus­tom­ers and shar­ing data di­git­ally, with an in­ter­face to core sys­tems and di­gital chan­nels. It was built with a vendor part­ner­ship an­nounced in May 2019 and is de­scribed as a hy­brid of cloud and dis­trib­uted ledger, draw­ing data from gov­ern­ment bod­ies and sup­port­ing screen­ing, doc­u­ment veri­fic­a­tion and au­then­tic­a­tion without a phys­ical visit. The first bank went live on 30 April 2019, while a formal cent­ral bank an­nounce­ment of elec­tronic know-your-cus­tomer ap­pears in re­port­ing dated Feb­ru­ary 2021; the two dates are in­con­sist­ent.

On fraud, syn­thetic media, proven­ance, cryp­to­graphic trans­ition, pen­al­ties and ma­chine iden­tity, noth­ing was loc­ated for Bahrain.

Oman and Kuwait: a usage figure and a mandate, and little else

Oman publishes an authentication volume but no user count, and Kuwait publishes a mandate but no circular. Between them they constitute the thinnest published record of the six.

The po­s­i­tion in Oman and Kuwait is that each has one hard, dated fact in the di­gital trust field and al­most no sup­port­ing re­cord. Oman's cent­ral bank is­sued in­struc­tions on di­gital on­board­ing and elec­tronic know-your-cus­tomer in June 2023, and the na­tional di­gital iden­tity is re­por­ted to have handled 202.9 mil­lion au­then­tic­a­tion re­quests by the end of 2025. Kuwait's cent­ral bank, in April 2026, re­quired li­censed ex­change com­pan­ies to au­then­tic­ate cus­tom­ers through the gov­ern­ment mo­bile iden­tity ap­plic­a­tion be­fore each on­line in­ter­na­tional trans­fer. Neither state pub­lishes an iden­tity user count, a fraud loss fig­ure, a proven­ance or deep­fake in­stru­ment, a cryp­to­graphic pro­gramme, or an en­force­ment ac­tion touch­ing iden­tity or fraud. It is to be noted that in both cases the strongest avail­able doc­u­ment is a news re­port of a reg­u­lator's de­cision rather than the de­cision it­self.

Oman

Oman's scheme is THEQA, a mo­bile ap­plic­a­tion de­scribed on the re­spons­ible min­istry's own di­gital trust page as an ad­vanced di­gital iden­tity which can be

used to ac­cess elec­tronic ser­vices and to sign doc­u­ments and trans­ac­tions elec­tron­ic­ally. It is op­er­ated through the na­tional com­puter emer­gency read­i­ness team within the Min­istry of Trans­port, Com­mu­nic­a­tions and In­form­a­tion Tech­no­logy, and the same page re­cords that it is avail­able to all cit­izens and res­id­ents, work­ing along­side the per­sonal iden­tity card and the res­id­ency card. En­rol­ment is de­scribed as re­quir­ing an iden­tity card read­able by the hand­set over near-field com­mu­nic­a­tion, with face match­ing and a one-time passcode.

Two fur­ther facts are worth sep­ar­at­ing by re­li­ab­il­ity. The Royal Oman Po­lice stated in Septem­ber 2025 that the di­gital iden­tity may be presen­ted law­fully to the com­pet­ent au­thor­it­ies where the phys­ical cards are not held, which con­verts the cre­den­tial from a con­veni­ence into a sub­sti­tute. The fig­ure of 202.9 mil­lion au­then­tic­a­tion re­quests by the end of 2025 is at­trib­uted to the gov­ern­ment's own di­gital trans­form­a­tion re­port but was re­trieved from a sec­ond­ary site. It is a volume fig­ure, not a pop­u­la­tion fig­ure, and the num­ber of re­quests can­not be con­ver­ted into the num­ber of hold­ers. A com­mer­cial use began in Novem­ber 2025, when a tele­com­mu­nic­a­tions op­er­ator al­lowed cus­tom­ers to verify iden­tity through the ap­plic­a­tion in place of hand­ing over a phys­ical card.

The 202.9 mil­lion fig­ure is at­trib­uted to the na­tional di­gital trans­form­a­tion re­port for 2025 but was re­trieved from a sec­ond­ary site. It counts au­then­tic­a­tion re­quests. It is not a user count and has not been con­ver­ted into one here.

The au­thor­ity that binds for fin­an­cial ser­vices is the Cent­ral Bank of Oman, and its own site re­cords that in June 2023 it is­sued in­struc­tions on di­gital on­board­ing and elec­tronic know-your-cus­tomer, set­ting out the cus­tomer cat­egor­ies and products covered. Re­port­ing at the time con­firmed that the in­struc­tions went to all banks and other li­censed in­sti­tu­tions. The veri­fic­a­tion back­bone named in the frame­work is Mala'a, the na­tional di­gital on­board­ing re­gistry, run by the credit and fin­an­cial in­form­a­tion centre under cent­ral bank su­per­vi­sion and in­ten­ded to serve as an in­de­pend­ent and re­li­able data source against which in­sti­tu­tions verify in­form­a­tion dur­ing di­gital on­board­ing. A second phase cov­er­ing busi­ness on­board­ing is at proof-of-concept stage with three par­ti­cip­at­ing mem­bers.

Digital TrustVolge Quarterly28 •

On data pro­tec­tion Oman has a law and an un­re­con­ciled pen­alty pic­ture. The Per­sonal Data Pro­tec­tion Law was is­sued as Royal De­cree 6 of 2022 on 9 Feb­ru­ary 2022; its ex­ec­ut­ive reg­u­la­tions took ef­fect on 5 Feb­ru­ary 2024 with a one-year grace period, and one legal up­date re­cords the law as be­com­ing fully en­force­able on 5 Feb­ru­ary 2026. The pen­alty fig­ures con­flict sharply across pro­fes­sional com­ment­ar­ies: one gives court-im­posed fines of five hun­dred to five hun­dred thou­sand rials with min­istry ad­min­is­trat­ive fines of up to two thou­sand rials, an­other gives twenty thou­sand to one hun­dred thou­sand rials for pro­cessing sens­it­ive data without a per­mit, and a third sets out three tiers. These are set out as con­flict­ing rather than av­er­aged. No en­force­ment ac­tion touch­ing iden­tity or fraud was loc­ated, and noth­ing was found on syn­thetic media, proven­ance, cryp­to­graphic trans­ition, or ma­chine and agent iden­tity.

Kuwait

Kuwait has made a national identity application a precondition of a financial transaction, which is further than any other Gulf state has gone in public.

Kuwait's scheme is the Kuwait Mo­bile ID ap­plic­a­tion, also rendered as My Iden­tity, is­sued by the Pub­lic Au­thor­ity for Civil In­form­a­tion and de­scribed by that au­thor­ity as a se­cure mo­bile di­gital iden­tity de­rived from the civil iden­tity card, sup­port­ing iden­tity veri­fic­a­tion, login to elec­tronic ser­vices and di­gital sig­na­ture. No re­gistered user fig­ure has been pub­lished.

The in­stru­ment that binds is the Cent­ral Bank of Kuwait's dir­ect­ive to li­censed ex­change com­pan­ies, re­por­ted on 22 and 23 April 2026. Its re­quire­ments as re­por­ted are threefold: cus­tom­ers must be au­then­tic­ated through the gov­ern­ment mo­bile iden­tity ap­plic­a­tion be­fore each on­line in­ter­na­tional trans­fer, firms must es­tab­lish dir­ect elec­tronic in­teg­ra­tion with the civil in­form­a­tion au­thor­ity, and re­cords of each veri­fic­a­tion step must be re­tained. Re­port­ing notes that the lar­ger ex­change com­pan­ies are already linked to the au­thor­ity while oth­ers must up­grade their sys­tems within a time­frame set by the cent­ral bank; that time­frame has not been pub­lished, and no ef­fect­ive date ap­peared in the cov­er­age re­viewed. A re­lated meas­ure re­duced the daily cash trans­ac­tion limit at ex­change com­pan­ies from three thou­sand dinars to one thou­sand, which moves lar­ger trans­fers to­wards banks.

The Kuwaiti dir­ect­ive is de­scribed from news re­port­ing dated 22 and 23 April 2026 and from a vendor sum­mary. The cent­ral bank cir­cu­lar it­self was not re­trieved, and no com­pli­ance dead­line was pub­lished in the cov­er­age re­viewed.

Bey­ond that dir­ect­ive the Kuwaiti re­cord is empty on every other head­ing in this sur­vey. One vendor guide states that the cent­ral bank has is­sued di­gital on­board­ing in­struc­tions and ap­proved elec­tronic know-your-cus­tomer gen­er­ally; that claim could not be veri­fied against a cent­ral bank text. No fraud loss fig­ure, no deep­fake or proven­ance meas­ure, no cryp­to­graphic pro­gramme, no pub­lished pen­alty touch­ing iden­tity or fraud, and noth­ing on ma­chine or agent iden­tity was loc­ated. Nor was a Kuwaiti data pro­tec­tion in­stru­ment re­trieved, which is a gap in the re­cord rather than a state­ment that none ex­ists.

What the two po­s­i­tions ask of a buyer is that the Kuwaiti pre­ced­ent be taken ser­i­ously and the Omani fig­ures be handled care­fully. Kuwait has made a state iden­tity ap­plic­a­tion a pre­con­di­tion of a par­tic­u­lar fin­an­cial trans­ac­tion, which is a stronger form of de­pend­ency than per­mit­ting a state cre­den­tial as one route among sev­eral, and an en­ter­prise op­er­at­ing re­mit­tance flows there now has a single point of fail­ure it did not choose. Oman of­fers a cre­den­tial with legal stand­ing as a sub­sti­tute for the phys­ical card and a pub­lished volume of use, but no basis on which to es­tim­ate cov­er­age. In both mar­kets the prac­tical step is to ob­tain the in­stru­ment it­self, in Ar­abic if ne­ces­sary, rather than to rely on the re­port­ing of it.

The six compared: one region, six different answers

Laid side by side, the six states share a pattern and almost nothing else. Every one has a national identity scheme and a financial-services instrument, and no two agree on what the credential is for, who may use it remotely, or what is disclosed about any of it.

What the table shows is that the Gulf is not a mar­ket in di­gital trust terms but six mar­kets which hap­pen to be ad­ja­cent. All six op­er­ate a na­tional di­gital iden­tity scheme and all six have a fin­an­cial-sec­tor in­stru­ment touch­ing re­mote on­board­ing or au­then­tic­a­tion. Bey­ond that the columns di­verge: two of the six pub­lish an iden­tity fig­ure that can be quoted at all, and both do so in forms that con­tra­dict them­selves; one pub­lishes a named au­then­tic­a­tion pro­hib­i­tion with a date; one makes a state ap­plic­a­tion a pre­con­di­tion of a trans­ac­tion; one re­stricts di­gital on­board­ing by the cus­tomer's coun­try of res­id­ence; and ex­actly one of the six pub­lishes en­force­ment fig­ures in a form a buyer could use to cal­ib­rate risk. It is to be noted that a veri­fic­a­tion flow built for any one of these six will

Digital TrustVolge Quarterly29 •

fail in at least one other, and that the fail­ure will not be tech­nical.

StateScheme and issuerPublished users or volumeFinancial-services instrumentAccepted remote routesPublished enforcement
Saudi ArabiaNafath, National Information Center, overseen by the data and AI authority32 million digital identities issued, stated 2026; 32 million beneficiaries, 1,245 platforms, 6.5 billion transactions reported at LEAP 2026; 23.5 million users and 3 billion verifications attributed to the authority, November 2024. UnreconciledCounter-Fraud Framework v1.0, October 2022, full compliance 29 June 2023; Counter-Fraud Fundamental Requirements, 14 April 2025, reported effective 13 April 2026; quantum resilience circular, 27 August 2026Government identity route against the national register; the account of Nafath for nationals and Absher for residents rests on secondary commentary only48 decisions under Article 36 of the data protection law, announced 16 January 2026; amounts not disclosed
United Arab EmiratesUAE PASS, with the telecommunications and digital government regulatorMore than 11 million registered users on the regulator's page; more than 12 million per a World Economic Forum piece; more than 14 million per a Digital Dubai post, July 2026. UnreconciledNotice 2025/3057, text-message and electronic-mail passcodes withdrawn for digital banking by 31 March 2026; Notice 2176/2026 on Android malware; Guidance on Digital Identification for Customer Due Diligence, 31 October 2022Digital identity systems permitted, including UAE PASS and the federal identity authority's validation gateway; Emirates identity document and verification record retained200 million dirhams on a foreign bank branch, 30 October 2026, with 500,000 and 300,000 dirhams on two officers; 20 million dirhams, 24 June 2026; approximately 222.92 million dirhams January to October 2026
QatarTawtheeq, the National Authentication System, transport and communications ministry; Qatar Digital ID, Ministry of InteriorNot publishedElectronic know-your-customer procedures issued 15 October 2023; Payment Services Regulation requires compliance with the eKYC security principlesMinistry of Interior identity card for citizens and residents, commercial registration for companies, passport for visitors and non-residentsNot published
StateScheme and issuerPublished users or volumeFinancial-services instrumentAccepted remote routesPublished enforcement
BahraineKey, the National Authentication Framework, Information and eGovernment AuthorityNot publishedFinancial crime module of the central bank rulebook, amendments recorded 2022 and 2023; capital markets digital onboarding added October 2023; national eKYC platform letter indexed 17 February 2025National eKYC system; digital onboarding barred for natural persons resident outside the Council save for Golden Visa holders and Council nationals holding a Bahraini personal number; proprietary identity applications require review and geolocationNot published
OmanTHEQA, national computer emergency readiness team, transport and communications ministry202.9 million authentication requests to end 2025, attributed to the national digital transformation report. No user count publishedInstructions on digital onboarding and electronic know-your-customer, June 2023Verification against Mala'a, the national digital onboarding registry, under central bank supervision; business onboarding at proof of concept with three membersNot published
KuwaitKuwait Mobile ID, Public Authority for Civil InformationNot publishedCentral bank directive to exchange companies, reported 22 and 23 April 2026; no published compliance dateAuthentication through the government mobile identity application before each online international transfer, with direct electronic integration to the civil information authorityNot published

Entries marked not pub­lished mean that no fig­ure or ac­tion was loc­ated in the sources re­trieved for this part, not that none ex­ists. Fig­ures in the users column are re­pro­duced in the form and with the met­ric their sources used; none has been con­ver­ted, roun­ded or com­bined. The Saudi and Emir­ati user fig­ures are in­tern­ally un­re­con­ciled and are presen­ted as such.

What the comparison shows

The first thing the table shows is that the cre­den­tial is the easy part. All six states have solved is­su­ance: a mo­bile ap­plic­a­tion tied to a na­tional re­gister, with a bio­met­ric or doc­u­ment-read­ing en­rol­ment step, is now the stand­ard shape of a Gulf di­gital iden­tity. What dif­fers is the legal weight at­tached to the cre­den­tial and the ob­lig­a­tions that sit around it. Oman has made

Digital TrustVolge Quarterly30 •

its ap­plic­a­tion a law­ful sub­sti­tute for the phys­ical card. Kuwait has made its ap­plic­a­tion a pre­con­di­tion of a trans­ac­tion. Bahrain has writ­ten the na­tional plat­form into its rule­book with named res­id­ency ex­cep­tions. The Emir­ates has per­mit­ted its scheme as one route among sev­eral while sep­ar­ately pro­hib­it­ing a com­pet­ing au­then­tic­a­tion method out­right. Saudi Ar­a­bia has built the deep­est set of sur­round­ing ob­lig­a­tions without pub­lish­ing a veri­fi­able reach for the cre­den­tial it­self. Qatar has an in­stru­ment and a cre­den­tial and no doc­u­mented bridge between them.

The second thing the table shows is that the bind­ing in­stru­ment is al­most never an iden­tity in­stru­ment. In five of the six cases the thing that ac­tu­ally con­strains an en­ter­prise is a fin­an­cial crime or fraud rule is­sued by the cent­ral bank: a counter-fraud frame­work, an anti-money-laun­der­ing guid­ance note, a fin­an­cial crime mod­ule, a set of on­board­ing in­struc­tions, a dir­ect­ive to ex­change com­pan­ies. Di­gital iden­tity enters as a per­mit­ted or re­quired route in­side that rule. The prac­tical con­sequence is that the au­thor­ity a buyer must sat­isfy on iden­tity is a pruden­tial reg­u­lator, and the evid­ence it will want is the evid­ence a pruden­tial reg­u­lator wants: doc­u­mented con­trols, re­tained re­cords, ma­tur­ity as­sess­ments and in­cid­ent no­ti­fic­a­tion. An iden­tity pro­gramme presen­ted as a tech­no­logy pro­ject will be ex­amined as a com­pli­ance pro­ject.

The credential is the easy part. What differs across the six is the weight the law attaches to it.

The third thing the table shows is an asym­metry of dis­clos­ure so large that it dis­torts per­cep­tion of the re­gion. One state pub­lishes en­force­ment amounts, down to in­di­vidual of­ficers, on a rolling basis: the Emir­ates re­cor­ded ap­prox­im­ately 222.92 mil­lion dirhams in pen­al­ties between Janu­ary and Oc­to­ber 2026 and 214.5 mil­lion in 2025. One pub­lishes a count of de­cisions without amounts: Saudi Ar­a­bia, forty-eight de­cisions an­nounced in Janu­ary 2026. Four pub­lish neither. That does not mean four Gulf states are not en­for­cing. It means that any com­par­at­ive state­ment about en­force­ment in­tens­ity in the Gulf is, at present, a state­ment about Emir­ati and Saudi dis­clos­ure prac­tice. The same ap­plies to ad­op­tion: a pub­lished claim that iden­tity ad­op­tion across the Coun­cil ranges from single di­gits to around eighty per cent of the adult pop­u­la­tion comes from a vendor guide, and the two lead­ers in that ac­count are pre­cisely the two states whose own fig­ures are in­tern­ally in­con­sist­ent.

The ad­op­tion range of single di­gits to around eighty per cent of the adult pop­u­la­tion, and the char­ac­ter­isa­tion of the Gulf as six dis­tinct reg­u­lat­ory and tech­nical en­vir­on­ments, come from a com­mer­cial veri­fic­a­tion vendor's guide re­por­ted in the trade press in Au­gust 2026. They are not of­fi­cial fig­ures.

The fourth thing the table shows is where the re­gional flows break. Bahrain's rule turns on the cus­tomer's coun­try of res­id­ence, not on the strength of the veri­fic­a­tion: a per­son res­id­ent out­side the Coun­cil can­not be on­boarded di­git­ally by a Bahraini li­censee un­less they hold a Golden Visa or are a Coun­cil na­tional with a Bahraini per­sonal num­ber. Kuwait's rule turns on a single ap­plic­a­tion in the hands of one au­thor­ity. The Emir­ates re­quires re­ten­tion of the Emir­ates iden­tity doc­u­ment along­side the di­gital veri­fic­a­tion re­cord. Qatar enu­mer­ates ac­cept­able doc­u­ments by cus­tomer class. Oman routes veri­fic­a­tion through a na­tional on­board­ing re­gistry rather than through the iden­tity cre­den­tial it­self. A single re­gional on­board­ing flow can­not sat­isfy those five con­di­tions sim­ul­tan­eously, and the dif­fer­ences are not in the bio­met­ric layer, where products are in­ter­change­able, but in eli­gib­il­ity, doc­u­ment re­ten­tion and per­mit­ted cre­den­tial, where they are not. It fol­lows that the in­teg­ra­tion risk in a Gulf iden­tity pro­gramme sits in the rule­book rather than in the tech­no­logy, and that it will not be found by test­ing the tech­no­logy.

What fol­lows for a buyer is a change in the unit of plan­ning. The unit is not the Gulf and it is not the coun­try: it is the com­bin­a­tion of a cus­tomer's res­id­ency class, the li­cence held by the en­tity on­board­ing them, and the cre­den­tial avail­able to them. Six na­tional schemes gen­er­ate con­sid­er­ably more than six such com­bin­a­tions, and the ones that fail do so at the point of reg­u­lat­ory re­view or at a volume peak rather than in test­ing. The second con­sequence is evid­en­tiary. Where four of six reg­u­lat­ors pub­lish no en­force­ment re­cord, the ab­sence of a pub­lic pen­alty can­not be read as tol­er­ance, and the prudent as­sump­tion is that the ob­lig­a­tions in the rule­book are the ob­lig­a­tions that will be ex­amined, whether or not any­one has yet been fined for breach­ing them. The prac­tical test of a Gulf iden­tity pro­gramme is there­fore not whether it veri­fies ac­cur­ately, which is now a solved prob­lem, but whether it can be shown to verify in the man­ner each reg­u­lator has spe­cified.

Take the Saudi po­s­i­tion first, be­cause it is the sharpest. A sur­vey of fraud man­age­ment, anti-money-laun­der­ing and com­pli­ance lead­ers at banks in the King­dom, pub­lished on 19 Au­gust 2026, found that eighty-five per cent of those lead­ers saw fraud at­tempts at their in­sti­tu­tion in­creas­ing every year and eighty-three per cent saw losses in­creas­ing. Half of those sur­veyed said their bank loses more than ten mil­lion dol­lars

Digital TrustVolge Quarterly31 •

an­nu­ally to fraud and scams, and twenty-two per cent re­por­ted an­nual losses ex­ceed­ing twenty-five mil­lion dol­lars. Fifty-three per cent said the in­vest­ig­at­ive pro­cess still re­lies on manual re­view, and thirty-six per cent said in­vest­ig­a­tions often take longer than two days, which the pub­lisher de­scribed as the slow­est pace of any coun­try it sur­veyed.

The generative fraud picture: vendor surveys standing in for public data

There is no official figure for fraud losses in any Gulf state. What circulates instead is self-reported survey data from fraud-prevention vendors, and one court-documented deepfake case that is now nearly seven years old.

The po­s­i­tion on gen­er­at­ive fraud in the Gulf is that the pub­lished evid­ence is weaker than the volume of com­ment­ary sug­gests, and that the gap is not ac­ci­dental. No cent­ral bank in the six states has pub­lished an ag­greg­ate fraud loss fig­ure. The num­bers in cir­cu­la­tion come al­most en­tirely from sur­veys com­mis­sioned by vendors of fraud-pre­ven­tion tech­no­logy, in which bank ex­ec­ut­ives self-re­port loss bands, and from press ac­counts of po­lice and min­istry warn­ings. The one case with a doc­u­mented fig­ure and a court fil­ing be­hind it con­cerns a voice clone used against a bank in the Emir­ates in Janu­ary 2020. It is to be noted that a buyer build­ing a busi­ness case on Gulf fraud losses is build­ing it on self-re­por­ted sur­vey bands, and that the or­gan­isa­tions which com­mis­sioned those sur­veys sell the rem­edy.

The Emir­ati po­s­i­tion comes from the same pub­lisher's re­gional edi­tion, re­leased in April 2026 and drawn from a sur­vey of one hun­dred fraud, anti-money-laun­der­ing and com­pli­ance lead­ers at banks in the coun­try, con­duc­ted between Decem­ber 2025 and Janu­ary 2026. Fifty-eight per cent re­por­ted rising fraud losses. Ninety-five per cent said they struggle to de­tect micro-scams ori­gin­at­ing on so­cial media. Eighty-three per cent non­ethe­less rated their own con­trols as ef­fect­ive or very ef­fect­ive. Twenty-six per cent said their or­gan­isa­tion re­im­burses more than half of scam vic­tims, which the pub­lisher char­ac­ter­ised as below global peers. A fig­ure of sixty-six per cent for rising at­tempts, and one of sixty-two per cent es­tim­at­ing an­nual losses above five mil­lion dol­lars, ap­pear in sec­ond­ary cov­er­age of the same work and could not be con­firmed in the pub­lisher's own re­lease.

Both sets of fig­ures above are self-re­por­ted sur­vey re­sponses from bank ex­ec­ut­ives, com­mis­sioned by a vendor of be­ha­vi­oural fraud-de­tec­tion tech­no­logy. They are not reg­u­lat­ory data and they are not audited. The sixty-six and sixty-two per cent fig­ures for the Emir­ates are from sec­ond­ary cov­er­age and are un­con­firmed.

What the regulators have and have not said

Against that, the reg­u­lat­ory re­cord is one of warn­ing rather than meas­ure­ment. In the Emir­ates, the Min­istry of Human Re­sources and Emir­at­isa­tion warned in Septem­ber 2026 that fraud­sters could use deep­fake tech­no­logy to cre­ate con­tent ap­pear­ing to ori­gin­ate from the min­istry or its of­fi­cials, and the Cyber Se­cur­ity Coun­cil has is­sued re­peated no­tices ur­ging veri­fic­a­tion of video and audio clips. In Qatar, the Min­istry of In­terior warned in Au­gust 2026 that ar­ti­fi­cially gen­er­ated con­tent was being used to im­per­son­ate in­di­vidu­als. A com­mer­cial threat in­tel­li­gence firm warned in the same month that deep­fake in­vest­ment schemes could spread into Gulf mar­kets. None of these car­ries a fig­ure.

Two fur­ther fig­ures cir­cu­late and should be handled with care. One ac­count at­trib­utes to pre­lim­in­ary cent­ral bank fig­ures in the Emir­ates a total of 120 mil­lion dirhams drained by hy­brid phish­ing and for­gery schemes in 2025, with syn­thetic audio and video play­ing a part. That fig­ure could not be loc­ated in any cent­ral bank pub­lic­a­tion and is not treated here as es­tab­lished. A second ac­count, from Au­gust 2026, re­ports doc­u­mented in­di­vidual losses of between 85,711 and 200,000 dirhams in cases where deep­fakes lent cred­ib­il­ity to un­li­censed trad­ing plat­forms; that range too rests on a single press ac­count. Neither should be used as a re­gional loss es­tim­ate.

The only Gulf deepfake fraud with a figure and a court filing behind it happened in January 2020.

The doc­u­mented case re­mains the one from Janu­ary 2020. A bank man­ager in the Emir­ates re­ceived a call from a voice he re­cog­nised as a dir­ector of a large com­pany with which he had pre­vi­ously dealt, re­quest­ing thirty-five mil­lion dol­lars for an ac­quis­i­tion, sup­por­ted by forged elec­tronic mail pur­port­ing to come from the com­pany and its law­yer. In­vest­ig­at­ors con­cluded that voice-clon­ing soft­ware had been used to re­pro­duce the dir­ector's voice. The case be­came pub­lic through a fil­ing by the United States De­part­ment of Justice for an ex parte order in the Dis­trict Court for the Dis­trict of Columbia, in con­nec­tion with a money laun­der­ing in­vest­ig­a­tion in the Emir­ates, after au­thor­it­ies there sought de­tails of two United States ac­counts into which more than four hun­dred thou­sand dol­lars of the pro­ceeds had been de­pos­ited. In­vest­ig­at­ors be­lieved at least sev­en­teen people were in­volved. Neither the bank nor the com­pany was named.

Digital TrustVolge Quarterly32 •

The ar­gu­ment that fol­lows is un­com­fort­able but it is the hon­est one. The Gulf has a well-doc­u­mented deep­fake-en­abled fraud with a fig­ure at­tached, and it is nearly seven years old. Everything since con­sists of warn­ings without fig­ures and fig­ures without reg­u­lat­ory stand­ing. This is not evid­ence that gen­er­at­ive fraud in the re­gion is small. It is evid­ence that nobody with the data is pub­lish­ing it. Saudi Ar­a­bia's Counter-Fraud Fun­da­mental Re­quire­ments, in force since April 2026, ob­lige in­sti­tu­tions to re­port the value of cus­tomer and op­er­a­tional fraud losses, and that re­port­ing is ac­cu­mu­lat­ing now. The Emir­ates re­views au­then­tic­a­tion com­pli­ance at six-monthly in­ter­vals and has shif­ted li­ab­il­ity for passcode-linked fraud onto the in­sti­tu­tion, which places the loss data with the in­sti­tu­tions by design. In both cases the num­bers exist and are not pub­lic.

There is also a meas­ure­ment prob­lem be­neath the dis­clos­ure prob­lem. The Saudi sur­vey re­cords in­sider fraud and syn­thetic iden­tity fraud tied as the threats ex­ec­ut­ives worry about most, and so­cial en­gin­eer­ing reach­ing cus­tom­ers through dat­ing and gam­ing ap­plic­a­tions rather than the tele­phone and elec­tronic mail chan­nels banks mon­itor. The Emir­ati sur­vey re­cords near-uni­ver­sal dif­fi­culty de­tect­ing micro-scams ori­gin­at­ing on so­cial media. Both are cat­egor­ies that fall out­side con­ven­tional trans­ac­tion mon­it­or­ing, so an in­sti­tu­tion re­port­ing its losses ac­cur­ately under the Saudi re­quire­ment may still be re­port­ing only the part of the prob­lem its sys­tems can see.

What should be done dif­fer­ently fol­lows from the shape of the evid­ence rather than its con­tent. In­ternal loss data, which the Saudi re­quire­ment now com­pels in­sti­tu­tions to pro­duce in any case, is the only Gulf fraud fig­ure that can be de­fen­ded line by line, and it should dis­place sur­vey bands in any in­vest­ment case put to a board. Where a re­gional es­tim­ate is un­avoid­able, it should be presen­ted with its proven­ance named, and the dis­tinc­tion between a reg­u­lator's fig­ure, a court fil­ing and a vendor sur­vey car­ried into the doc­u­ment rather than flattened out of it. And de­tec­tion scope should be ex­amined be­fore de­tec­tion tech­no­logy: a con­trol that can­not see a scam which began on a dat­ing ap­plic­a­tion will not be im­proved by being made faster.

The ex­cep­tion is in­struct­ive pre­cisely be­cause it is so com­plete. On 11 May 2026 the Cyber Se­cur­ity Coun­cil of the Emir­ates signed an agree­ment with the com­mer­cial arm of the Ad­vanced Tech­no­logy Re­search Coun­cil to de­ploy and scale a defined set of com­pon­ents in sup­port of the na­tional post-quantum mi­gra­tion. The com­pon­ents are named: a crypto dis­cov­ery

tool for cryp­to­graphic dis­cov­ery, in­vent­ory man­age­ment and vul­ner­ab­il­ity as­sess­ment, cus­tom­ised to re­quire­ments set by the Na­tional Cryp­to­graphy Cen­ter; en­tan­gle­ment-based quantum key dis­tri­bu­tion, de­veloped at the na­tional re­search in­sti­tute and already used to con­nect data centres at one Abu Dhabi en­tity; and na­tional cryp­to­graphic lib­rar­ies de­veloped by that in­sti­tute, com­mer­cial­ised as the of­fi­cial cryp­to­graphic found­a­tion and con­tain­ing both clas­sical and post-quantum al­gorithms. A se­cur­ity labor­at­ory, de­scribed as one of the only fa­cil­it­ies in the coun­try ded­ic­ated to cryp­to­graphic com­pli­ance test­ing, val­id­ates cryp­to­graphic as­sets against na­tional reg­u­lat­ory re­quire­ments.

What is actually being procured, and what cannot be seen

Only one named, dated digital trust procurement in the Gulf could be established from public sources, and it is cryptographic rather than identity-related. The procurement that must be happening is almost entirely invisible.

The po­s­i­tion on Gulf di­gital trust pro­cure­ment is that it can­not be seen, and that the in­vis­ib­il­ity is struc­tural rather than in­cid­ental. Across six states, each with a na­tional iden­tity scheme and a fin­an­cial-sec­tor in­stru­ment re­quir­ing veri­fic­a­tion, au­then­tic­a­tion or fraud con­trol, one named and dated sup­plier se­lec­tion of con­sequence could be es­tab­lished from pub­lic sources dur­ing this work, and its sub­ject is cryp­to­graphy rather than iden­tity. No frame­work agree­ment for iden­tity veri­fic­a­tion, au­then­tic­a­tion or fraud tech­no­logy cov­er­ing a Gulf gov­ern­ment or a Gulf fin­an­cial sec­tor was loc­ated. No award no­tice nam­ing a sup­plier for such work was loc­ated. It is to be noted that a buyer at­tempt­ing to bench­mark price, scope or sup­plier per­form­ance against re­gional peers has, at present, al­most noth­ing to bench­mark against.

What makes that an­nounce­ment un­usual is not the money, which is not dis­closed, but the spe­cificity: the buyer, the sup­plier, the com­pon­ents, the stand­ards body whose re­quire­ments shaped one of them, and the date are all named. Noth­ing com­par­able was found on the iden­tity side in any of the six states. That asym­metry is it­self the find­ing: the Emir­ates chose to pub­lish a cryp­to­graphic se­lec­tion in de­tail while pub­lish­ing noth­ing equi­val­ent about the iden­tity and au­then­tic­a­tion es­tate on which its passcode with­drawal rule de­pends.

Bey­ond that, the search re­turns in­sti­tu­tions rather than trans­ac­tions. Saudi gov­ern­ment pro­cure­ment runs

Digital TrustVolge Quarterly33 •

through a cent­ral plat­form op­er­ated by the Min­istry of Fin­ance, cov­er­ing tenders, con­tract man­age­ment, budget­ing and pay­ments; re­gis­tra­tion on it con­fers no qual­i­fic­a­tion and no award. Com­ment­ary on for­eign sup­pli­ers de­scribes an ex­emp­tion route an­nounced in Feb­ru­ary 2026 with con­tracts up to one mil­lion riy­als ex­empt, a de­tail that could not be veri­fied against an of­fi­cial source. Tender ag­greg­at­ors re­turned iden­tity-veri­fic­a­tion no­tices for buy­ers out­side the Coun­cil. The mech­an­ism through which a Gulf state buys iden­tity tech­no­logy is vis­ible; the pur­chases are not.

Where supplier names do surface

Sup­plier names do ap­pear in the Gulf iden­tity re­cord, but through vendor an­nounce­ments rather than pro­cure­ment no­tices, and chiefly in Bahrain. The na­tional elec­tronic know-your-cus­tomer plat­form man­dated by the cent­ral bank and op­er­ated by the na­tional pay­ments com­pany was built with a tech­no­logy part­ner an­nounced in May 2019, de­scribed at the time as cre­at­ing the first na­tional know-your-cus­tomer util­ity in­cor­por­at­ing dis­trib­uted ledger tech­no­logy, with a first phase re­por­ted as de­ployed in four weeks and bio­met­ric veri­fic­a­tion linked to na­tional iden­tity card data. The busi­ness-fa­cing iden­tity and know-your-cus­tomer ser­vice de­rived from the na­tional eKey is de­livered through a com­mer­cial sub­si­di­ary of a Bahraini tele­com­mu­nic­a­tions group, and the Na­tional Bank of Bahrain is re­por­ted to have de­ployed it for pass­word­less di­gital iden­tity in Decem­ber.

The Bahraini sup­plier in­form­a­tion in the pre­ced­ing para­graph comes from vendor press re­leases and trade cov­er­age, not from pro­cure­ment re­cords. Con­tract val­ues, dur­a­tions and scopes are not pub­lished in any of those sources.

The mechanism by which a Gulf state buys identity technology is visible. The purchases are not.

What the invisibility costs

The ar­gu­ment is that this opa­city has three spe­cific costs to an en­ter­prise buyer, none of which is about trans­par­ency as a vir­tue. The first is pri­cing: without award val­ues, scopes or dur­a­tions there is no re­gional ref­er­ence point for what a na­tional-scale veri­fic­a­tion in­teg­ra­tion or a fraud plat­form costs, and a sup­plier pro­pos­ing a Gulf price can be chal­lenged only by ref­er­ence to mar­kets whose reg­u­lat­ory con­di­tions dif­fer. The second is ref­er­ence: a buyer can­not es­tab­lish from pub­lic sources which sup­pli­ers have op­er­ated at scale against which Gulf reg­u­lator's re­quire­ments, so

claims of re­gional ex­per­i­ence are dif­fi­cult to test and easy to make. The third is de­pend­ency: where a na­tional scheme is built on named com­pon­ents, the res­ult­ing con­cen­tra­tion is a fact about the buyer's own risk pro­file, and it is being es­tab­lished through vendor an­nounce­ments rather than dis­closed through pro­cure­ment.

There is a de­mand sig­nal in the sur­vey ma­ter­ial, and it should be read with the cau­tion that at­taches to its source. The Saudi bank sur­vey pub­lished in Au­gust 2026 found half of re­spond­ents say­ing their in­sti­tu­tion was act­ively pro­cur­ing or im­ple­ment­ing new fraud-pre­ven­tion tool­ing, which its pub­lisher re­por­ted as the highest pro­por­tion of any coun­try it sur­veyed and twenty-seven per­cent­age points above its own global av­er­age. If that pro­por­tion is any­where near ac­cur­ate, a sub­stan­tial volume of fraud and iden­tity pro­cure­ment is under way in the King­dom and es­sen­tially none of it is vis­ible in pub­lic re­cords. The two find­ings are con­sist­ent: Gulf fin­an­cial in­sti­tu­tions are private or state-linked buy­ers pur­chas­ing under com­mer­cial terms, and neither they nor their sup­pli­ers have an in­cent­ive to pub­lish.

A fourth cat­egory de­serves sep­ar­ate treat­ment be­cause it is the one most likely to be mis­taken for pro­cure­ment in­tel­li­gence. Vendor guides, in­teg­ra­tion hand­books and coun­try ex­plain­ers now make up the bulk of Eng­lish-lan­guage writ­ing on Gulf iden­tity veri­fic­a­tion, and they are fre­quently the only source for op­er­a­tion­ally sig­ni­fic­ant claims: which ac­count tiers may be opened re­motely in Saudi Ar­a­bia, whether act­ive live­ness is re­quired in Qatar, whether elec­tronic know-your-cus­tomer is ap­proved gen­er­ally in Kuwait. Those doc­u­ments are mar­ket­ing, and sev­eral of the claims they carry could not be traced to any reg­u­lator's text. They are a map of what sup­pli­ers be­lieve buy­ers want to hear, not a sub­sti­tute for the in­stru­ment.

Digital TrustVolge Quarterly34 •

What should be done dif­fer­ently is to treat the ab­sence of com­par­ables as a ne­go­ti­at­ing con­di­tion rather than as an in­con­veni­ence. Where no re­gional price ref­er­ence ex­ists, scope and exit terms carry the weight that bench­mark­ing would oth­er­wise carry, and con­tracts should be writ­ten so that a sup­plier's per­form­ance against a spe­cific reg­u­lator's re­quire­ment is demon­strable rather than as­ser­ted. Where a sup­plier claims Gulf ex­per­i­ence, the ap­pro­pri­ate re­quest is the reg­u­lator, the in­stru­ment and the date, since those are the facts that would ap­pear in a pro­cure­ment re­cord if one were pub­lished. And where a na­tional scheme's com­pon­ents are known, the res­ult­ing con­cen­tra­tion should be entered in the en­ter­prise's own risk re­gister, be­cause no one else is going to enter it.

The machine identity gap across the six schemes

Not one of the six Gulf states has published anything on the identity of a machine or an autonomous software agent. Every national scheme in the region assumes that the thing being authenticated is a person.

The po­s­i­tion on ma­chine and agent iden­tity in the Gulf is that there is no po­s­i­tion. Across Saudi Ar­a­bia, the United Arab Emir­ates, Qatar, Bahrain, Oman and Kuwait, no in­stru­ment, guideline, con­sulta­tion paper or pub­lished state­ment ad­dress­ing the iden­tity of a ma­chine, a ser­vice or an autonom­ous soft­ware agent was loc­ated dur­ing this work. Every one of the six na­tional schemes is built on a single premise: that the en­tity present­ing it­self is a nat­ural per­son, hold­ing a na­tional iden­tity re­cord, whose live­ness and bio­met­ric match can be tested. It is to be noted that this is now the largest single gap in the Gulf di­gital trust fab­ric, that it is uni­form across all six states, and that other jur­is­dic­tions have begun to close it, which means the gap can be meas­ured rather than merely as­ser­ted.

Con­sider what each scheme ac­tu­ally au­then­tic­ates. Na­fath per­forms a real-time match against the Saudi na­tional iden­tity or res­id­ency re­cord. UAE PASS per­forms a face or fin­ger­print check against the Emir­ates iden­tity re­cord held in gov­ern­ment sys­tems. Oman's THEQA reads the iden­tity card over near-field com­mu­nic­a­tion and per­forms face match­ing with a passcode. Bahrain's eKey dis­tin­guishes a stand­ard ac­count, re­quir­ing only a per­sonal num­ber and pass­word, from an ad­vanced ac­count, re­quir­ing fin­ger­print veri­fic­a­tion in per­son. Kuwait's mo­bile iden­tity is de­rived from the civil iden­tity card. Qatar's na­tional au­then­tic­a­tion sys­tem sup­ports a pass­word with a second factor or a smart iden­tity card. In each case the

strength of the cre­den­tial de­rives from a bio­lo­gical fact about a human being and a re­cord held by a state about that human being. Neither is avail­able for a soft­ware agent.

The same as­sump­tion runs through the fin­an­cial-sec­tor in­stru­ments. The Emir­ati au­then­tic­a­tion no­tice with­draws passcodes de­livered to a per­son and sub­sti­tutes bio­met­ric checks, in-ap­plic­a­tion ap­proval, tokens and pass­keys, all of which pre­sup­pose a human at a device. The Kuwaiti dir­ect­ive re­quires that the cus­tomer be au­then­tic­ated through the gov­ern­ment iden­tity ap­plic­a­tion be­fore each on­line in­ter­na­tional trans­fer, which re­quires that a spe­cific per­son ap­prove a spe­cific pay­ment. Bahrain's rule­book con­di­tions di­gital on­board­ing on the cus­tomer's coun­try of res­id­ence and re­quires geo­loca­tion fea­tures in a pro­pri­et­ary iden­tity ap­plic­a­tion, both facts about a per­son's phys­ical po­s­i­tion.

The measurable distance

What makes the Gulf si­lence a gap rather than simply an early stage is that the com­par­ison now ex­ists. Singa­pore's media and in­focomm reg­u­lator launched a Model AI Gov­ernance Frame­work for Agen­tic AI in Janu­ary 2026 and re­is­sued it in May 2026 with case stud­ies, treat­ing an agent's iden­tity as some­thing to be unique, cata­logued and fur­nished with au­thor­isa­tions that are least-priv­ilege and non-trans­fer­able. In the United States, the Na­tional Cy­ber­se­cur­ity Cen­ter of Ex­cel­lence re­leased a concept paper in Feb­ru­ary 2026 on ap­ply­ing ex­ist­ing iden­tity stand­ards to soft­ware and ar­ti­fi­cial in­tel­li­gence agents, with a com­ment period that closed in April 2026. None of the Gulf au­thor­it­ies sur­veyed here has is­sued an equi­val­ent.

The Singa­por­ean and United States doc­u­ments are cited as the com­par­ison that makes the Gulf po­s­i­tion meas­ur­able. The in­dustry re­com­mend­a­tions re­ferred to are prac­ti­tioner guid­ance from a cloud se­cur­ity body, not reg­u­la­tion. No Gulf equi­val­ent of any of the three was loc­ated.

The prac­tical ex­pos­ure sits in three places, and it is worth being pre­cise about them. The first is au­thor­isa­tion: where a reg­u­lator re­quires that a cus­tomer ap­prove a trans­ac­tion through a na­tional iden­tity ap­plic­a­tion, an en­ter­prise per­mit­ting an auto­mated pro­cess to ini­ti­ate or ap­prove that trans­ac­tion has no doc­u­mented basis for doing so, be­cause the rule con­tem­plates a per­son and of­fers no al­tern­at­ive. The second is at­tri­bu­tion: where an in­stru­ment shifts li­ab­il­ity for au­then­tic­a­tion-linked fraud onto the in­sti­tu­tion, the in­sti­tu­tion must be able to say which actor au­thor­ised a dis­puted in­struc­tion, and if that actor was a soft­ware agent hold­ing a shared cre­den­tial it can­not. The third is on­board­ing: where re­mote on­board­ing rests on live­ness and bio­met­ric match, the con­trols exist to es­tab­lish

Digital TrustVolge Quarterly35 •

that a human is present, and an agen­tic flow de­feats the con­trol rather than fail­ing it.

There is a second com­par­ison worth mak­ing, which is in­ternal. The Gulf has shown that it can move quickly on a cryp­to­graphic ques­tion when it chooses to. Saudi Ar­a­bia's cent­ral bank set a cryp­to­graphic asset in­vent­ory dead­line for the end of 2026 and an en­ter­prise quantum risk as­sess­ment dead­line for the first quarter of 2027. The Emir­ates ap­proved a na­tional en­cryp­tion policy, stood up a na­tional cryp­to­graphy centre, and pro­cured dis­cov­ery tool­ing, key dis­tri­bu­tion and cryp­to­graphic lib­rar­ies under a dated agree­ment. Both states there­fore demon­strated, within a single eight­een-month win­dow, that they can identify an emer­ging cryp­to­graphic risk, name a pro­gramme, set dead­lines and buy com­pon­ents. The ab­sence of an equi­val­ent re­sponse on agent iden­tity is con­sequently not a cap­ab­il­ity con­straint. It is a ques­tion that has not yet been asked.

The Gulf closed the quantum question with deadlines and named components. On agent identity it has not yet asked the question.

Com­mer­cial act­ors have reached this con­clu­sion ahead of reg­u­lat­ors, which is worth re­cord­ing for what it in­dic­ates rather than for what it proves. The fraud-pre­ven­tion vendor whose Gulf sur­veys sup­ply most of the re­gion's cir­cu­lat­ing loss fig­ures has added agen­tic ac­count open­ing and agen­tic ac­count takeover to its own pub­lished use cases, and lists agent usage among the sig­nals it col­lects from bank­ing ses­sions. The same pub­lisher has is­sued sur­vey work whose title re­cords that bank­ing lead­ers in­ter­na­tion­ally see ar­ti­fi­cial in­tel­li­gence agents as the in­dustry's greatest vul­ner­ab­il­ity in the com­ing year. Both are com­mer­cial po­s­i­tions rather than evid­ence of in­cid­ence.

The two ob­ser­va­tions in the pre­ced­ing para­graph come from a fraud-pre­ven­tion vendor's own product pages and the title of its own sur­vey re­lease. They are com­mer­cial ma­ter­ial. No in­cid­ence fig­ure for agent-en­abled fraud in the Gulf has been pub­lished by any source loc­ated here.

What should be done dif­fer­ently does not re­quire wait­ing for a rule. The first step is an in­vent­ory, and the Saudi quantum cir­cu­lar has already es­tab­lished the prin­ciple that a cryp­to­graphic asset in­vent­ory is an audit­able arte­fact: the same dis­cip­line ap­plied to non-human cre­den­tials would es­tab­lish how many ser­vice ac­counts, keys and auto­ma­tion iden­tit­ies an en­ter­prise holds, which of them can ini­ti­ate a pay­ment, and which of them are shared. The second is to make at­tri­bu­tion pos­sible be­fore it is re­quired, by en­sur­ing that any auto­mated ac­tion against a cus­tomer re­cord re­solves to a dis­tinct, non-shared cre­den­tial with a re­cor­ded au­thor­isa­tion. The third is an­ti­cip­at­ory: the likely shape of a fu­ture re­quire­ment is vis­ible in the Singa­por­ean fram­ing of unique, cata­logued, least-priv­ilege and non-trans­fer­able au­thor­isa­tion, and an en­ter­prise that has already in­vent­or­ied and sep­ar­ated its non-human cre­den­tials will be an­swer­ing a ques­tion it has pre­pared for.

The Calls: 2027

Ten dated, falsifiable judgements on the year in digital trust, each with a confidence band and a stated reason, written to be graded in the 2028 Scorecard.

The ten Calls at a glance

Ten Calls for 2027, each dated and falsifiable, ordered from the most to the least certain; the full reasoning follows on the pages after.

Each Call is writ­ten to be graded (hit, miss or par­tial) in the 2028 Score­card. The con­fid­ence band is it­self a claim: High means the evid­ence base is meas­ured and the re­sid­ual un­cer­tainty is be­ha­viour, not dir­ec­tion; Me­dium means the dir­ec­tion is clear but the threshold is de­mand­ing or the tim­ing tight.

  • Call 1: GenAI fraud be­comes a board line item. High. Re­solves 2027
  • Call 2: Agen­tic iden­tity be­comes a pro­cure­ment re­quire­ment. Me­dium–High. Re­solves 2027
  • Call 3: OWASP Agen­tic Top 10 be­comes a re­view ar­ti­fact. Me­dium. Re­solves 2027
  • Call 4: The EU wal­let shifts from is­sued to re­quired. Me­dium. Re­solves 2027
  • Call 5: Over 40% of agen­tic pro­jects are can­celled. Me­dium–High. Re­solves 2027
  • Call 6: Con­tent proven­ance crosses to de­fault. Me­dium. Re­solves 2027
  • Call 7: Pass­keys over­take pass­words at a top-tier plat­form. Me­dium. Re­solves 2027
Digital TrustVolge Quarterly36 •
  • Call 8: Guard­ian-agent spend be­comes a fun­ded line. Me­dium. Re­solves 2027
  • Call 9: Gulf iden­tity rails reach agent pi­lots. Me­dium. Re­solves 2027
  • Call 10: India's DPDP cre­ates a live com­pli­ance mar­ket. Me­dium–High. Re­solves 2027

Reading the bands

Four Calls sit at Me­dium–High, five at Me­dium and one at High. The dis­tri­bu­tion is de­lib­er­ate: a for­ward-look that graded everything High would be mark­ing easy tar­gets. The single High Call (the fraud tra­ject­ory) is the one whose evid­ence base is a meas­ured, cor­rob­or­ated trend; everything down­stream of it car­ries the added un­cer­tainty of in­sti­tu­tional and mar­ket be­ha­viour.

The Calls

Each Call states a claim, a confidence band and the reasoning that supports it. The claims are written to be graded, not admired.

1. GenAI fraud becomes a board line item

United States gen­er­at­ive-AI-en­abled fraud losses con­tinue their track to­ward De­loitte's US$40bn 2027 fig­ure, and deep­fake fraud is treated as a board-level line item (named in risk re­gisters and pro­vi­sioned against) at a ma­jor­ity of large reg­u­lated in­sti­tu­tions by end-2027.

Con­fid­ence: High. Re­solves end-2027. Why this Call: The trend is meas­ured and steep, US$12.3bn (2023) to a US$40bn (2027) fore­cast at 32% (De­loitte, May 2024), and cor­rob­or­ated by En­trust's deep­fake-every-five-minutes and Sum­sub's ten­fold rise. The re­sid­ual un­cer­tainty is board be­ha­viour, not the loss tra­ject­ory, which is why con­fid­ence is High.

2. Agentic identity becomes a procurement requirement

First-class iden­tity for AI agents be­comes a named pro­cure­ment re­quire­ment for reg­u­lated buy­ers by end-2027, anchored on OAuth-for-agents and MCP En­ter­prise-Man­aged Au­thor­iz­a­tion (spe­cific­a­tion 2026-07-28) rather than be­spoke schemes.

Con­fid­ence: Me­dium–High. Re­solves end-2027. Why this Call: The plumb­ing landed in late 2025 (OpenID's agen­tic-iden­tity paper (Oct 2025), MCP OAuth 2.1 ex­ten­sions (Nov 2025), Okta Cross App Ac­cess into MCP), and the 109:1 ma­chine-to-human ratio (Palo Alto Net­works, 2026) makes the de­mand struc­tural. The risk is tim­ing: pro­cure­ment lan­guage lags cap­ab­il­ity by a year or more.

Con­fid­ence: Me­dium. Re­solves end-2027. Why this Call: OWASP's Top 10 for Agen­tic Ap­plic­a­tions v1.0

shipped 9–10 Dec 2025 with NIST, EC and Tur­ing re­view, and AIVSS sup­plies scor­ing. Ad­op­tion of a frame­work as a re­view norm typ­ic­ally takes longer than a single year, which caps con­fid­ence at Me­dium.

3. OWASP Agentic Top 10 becomes a review artifact

An OWASP Agen­tic Top 10 or AIVSS map­ping be­comes a stand­ard se­cur­ity-re­view ar­ti­fact for agen­tic sys­tems in reg­u­lated sec­tors by end-2027: ex­pec­ted in a re­view the way an OWASP web map­ping is today.

4. The EU wallet shifts from issued to required

The EU Di­gital Iden­tity Wal­let shifts from ‘is­sued’ to ‘must-be-ac­cep­ted’ dur­ing 2027 as the private-sec­tor ac­cept­ance duty binds, for­cing re­ly­ing parties in reg­u­lated sec­tors to re-ar­chi­tect iden­tity in­take.

Con­fid­ence: Me­dium. Re­solves end-2027. Why this Call: Reg (EU) 2024/1183 is in force (20 May 2024); wal­lets are due by 24 Decem­ber 2026 and ac­cept­ance late 2027. The legal clock is fixed, but the sup­ply side is slip­ping: as at 12 Oc­to­ber 2026 no mem­ber state held a cer­ti­fied wal­let, Ger­many is set for early Janu­ary 2027 and Sweden for 2029. The Call is there­fore graded on cer­ti­fied avail­ab­il­ity rather than on the ob­lig­a­tion in law, and the con­fid­ence band has been cut ac­cord­ingly.

5. Over 40% of agentic projects are cancelled

More than 40 per cent of agen­tic-AI pro­jects are can­celled by end-2027, with weak iden­tity and trust con­trols a lead­ing cause rather than model cap­ab­il­ity.

Con­fid­ence: Me­dium–High. Re­solves end-2027. Why this Call: This is Gart­ner's pub­lished fore­cast (2025), and the firm's judge­ment that only ~130 of thou­sands of ‘agen­tic’ vendors are real sup­ports the can­cel­la­tion rate. The falsifi­able half (that trust con­trols, not cap­ab­il­ity, lead the causes) is the harder claim, hence Me­dium–High.

6. Content provenance crosses to default

Con­tent proven­ance (C2PA / Con­tent Cre­den­tials) crosses from opt-in to in­fra­struc­ture de­fault across at least one major plat­form or CDN and one cam­era line by 2027.

Con­fid­ence: Me­dium. Re­solves end-2027. Why this Call: The pieces are mov­ing, Cloud­flare pre­serving on ~20% of the web (Feb 2025), in-cam­era sign­ing by Leica, Nikon, Sony and Sam­sung, and AI-Act Art­icle 50 (Aug 2026), but ‘de­fault’ is a high bar and plat­form com­mit­ments can slip, so Me­dium.

Digital TrustVolge Quarterly37 •

7. Passkeys overtake passwords at a top-tier platform

Pass­keys over­take pass­words as the de­fault au­then­tic­a­tion method for a named top-tier con­sumer co­hort by end-2027. En­ter­prise de­faults do not count: Mi­crosoft made pass­keys the de­fault sign-in in pub­lic-cloud Entra ID from 1 Septem­ber 2026, which this Call de­lib­er­ately ex­cludes.

Con­fid­ence: Me­dium. Re­solves end-2027. Why this Call: The pass­key/FIDO mar­ket com­pounds at 18.6% (Mar­ket.us, 2025) and the major plat­forms have shipped pass­keys, and the FIDO Al­li­ance re­ports more than 15 bil­lion pass­key-cap­able ac­counts against about 5 bil­lion in act­ive use, with 57 per cent of de­ploy­ing or­gan­isa­tions still re­ly­ing on phish­able primary meth­ods (State of Pass­keys 2026). Mak­ing pass­word­less the de­fault for a con­sumer co­hort, rather than merely avail­able, is a product de­cision car­ry­ing sup­port-cost fric­tion, so Me­dium.

8. Guardian-agent spend becomes a funded line

Guard­ian-agent and agent-over­sight spend­ing in­flects from a round­ing error into a fun­ded se­cur­ity line item dur­ing 2027, on the path to Gart­ner's pro­jec­ted 5–7 per cent of agen­tic spend by 2028.

Con­fid­ence: Me­dium. Re­solves end-2027. Why this Call: Gart­ner pro­jects guard­ian agents at 10–15% of the agen­tic mar­ket by 2030 and budgets rising from <1% (2026) to­ward 5–7% (2028); 2027 is the in­flec­tion year on that path. Early-mar­ket spend is volat­ile, so Me­dium.

9. Gulf identity rails reach agent pilots

In the Gulf, na­tional di­gital-iden­tity rails (Na­fath, UAE Pass) ex­tend into agent or en­ter­prise au­thor­isa­tion pi­lots by end-2027, bey­ond cit­izen au­then­tic­a­tion.

Con­fid­ence: Me­dium. Re­solves end-2027. Why this Call: The sov­er­eign schemes are ma­ture for cit­izens (Na­fath ~21.9M users) and the re­gional cyber en­vel­ope grows at 14.55% (Mordor, 2025), giv­ing both cap­ab­il­ity and ap­pet­ite. Ex­ten­sion to agents is plaus­ible but un­proven, so Me­dium.

10. India's DPDP creates a live compliance market

India's DPDP ob­lig­a­tions (Sig­ni­fic­ant Data Fi­du­ciary du­ties from 13 May 2027) turn iden­tity, con­sent-man­age­ment and audit into a live com­pli­ance mar­ket through 2027.

Con­fid­ence: Me­dium–High. Re­solves end-2027. Why this Call: The date is fixed in the Act and the In­dian di­gital-iden­tity mar­ket already com­pounds at 19.2% (GVR, 2026); en­force­ment forces de­mand for con­sent and audit tool­ing. The risk is the fa­mil­iar one of In­dian rule-mak­ing (phas­ing and rule de­tail), hence Me­dium–High rather than High.

How the Calls are made

A Call is not a forecast dressed as a fact; it is a dated, falsifiable judgement with a stated confidence and a stated reason, written so that it can be marked right or wrong in a year.

Volge Quarterly makes its method ex­pli­cit be­cause a for­ward-look that hides its reas­on­ing can­not be graded, and a for­ward-look that can­not be graded is a bro­chure. Each Call is built in four steps. First, it is anchored to a meas­ured trend or a fixed date (the De­loitte fraud tra­ject­ory, the eIDAS ac­cept­ance duty, the DPDP en­force­ment date) so that it rests on evid­ence rather than in­tu­ition. Second, it is writ­ten as a threshold that can be tested: not “agen­tic iden­tity will mat­ter” but “agen­tic iden­tity be­comes a named pro­cure­ment re­quire­ment for reg­u­lated buy­ers by end-2027.” Third, it is as­signed a con­fid­ence band that re­flects the re­sid­ual un­cer­tainty once the evid­ence is ac­coun­ted for. Fourth, its reas­on­ing is stated in full, so that a reader who dis­agrees knows pre­cisely which premise to at­tack.

The confidence bands

The bands are or­dinal, not prob­ab­il­istic; they rank the Calls by how much of the out­come is already de­term­ined by evid­ence versus how much de­pends on be­ha­viour still to come.

BandMeaningResidual uncertainty
HighEvidence base is measured and corroboratedInstitutional behaviour, not direction
Medium–HighDirection is clear and datedTiming, phasing, or a demanding second clause
MediumDirection is likelyA high threshold (‘default’, ‘standard’) or an unproven step
A forward-look that refuses to mark its own homework is just a brochure.

Two dis­cip­lines guard the method. Where mar­ket houses di­verge, the di­ver­gence is re­por­ted rather than av­er­aged, be­cause an av­er­age con­ceals the very dis­agree­ment that sig­nals an un­settled mar­ket. The

Digital TrustVolge Quarterly38 •

seven-point spread in di­gital-iden­tity CAGR is shown, not smoothed. And every fig­ure is year-stamped, so that a reader in 2028 can see not only what was claimed but on what basis it was claimed. The 2028 Score­card will grade each of these ten Calls hit, miss or par­tial, with a run­ning hit-rate re­por­ted by con­fid­ence band, so that the method it­self is held to ac­count and not merely the in­di­vidual judge­ments.

The year by domain

The same year read from four angles (technology, markets, regulation and delivery) and stress-tested against three scenarios for how 2027 could run.

The view from each angle

Read from four angles, 2027 tells one story in four registers: the technology consolidates, the market reallocates, the regulation binds and the delivery gap widens.

Technology

The tech­nical story of the year is con­ver­gence on stand­ards. Zero-trust ac­quires a ref­er­ence im­ple­ment­a­tion (NIST SP 1800-35, June 2025); agent iden­tity ac­quires a pro­tocol (MCP OAuth 2.1, ex­ten­ded Novem­ber 2025); con­tent ac­quires proven­ance (C2PA v2.2, May 2025); and cryp­to­graphy be­gins its post-quantum mi­gra­tion (NIST FIPS 203–205, Au­gust 2024). The com­mon thread is that trust prim­it­ives are being stand­ard­ised, which is the pre­con­di­tion for build­ing them in rather than bolt­ing them on.

Markets

The mar­ket story is real­loc­a­tion within a grow­ing en­vel­ope. In­form­a­tion-se­cur­ity spend rises to­ward US$247.5bn in 2026 (Gart­ner, 3Q26), and in­side it the iden­tity-cent­ric lay­ers (di­gital iden­tity at 20 per cent, zero-trust at 16.6 per cent, deep­fake de­tec­tion at 47.6 per cent) grow faster than the whole, tak­ing share from peri­meter de­fence. Cap­ital is fol­low­ing the ar­chi­tec­ture.

Regulation

The reg­u­lat­ory story is that the cal­en­dar fi­nally binds. Three dated ob­lig­a­tions, EU wal­let ac­cept­ance (late 2027), AI-Act trans­par­ency (Au­gust 2026) and DPDP en­force­ment (13 May 2027), land in­side four­teen months, con­vert­ing trust from a dis­cre­tion­ary pos­ture into a com­pli­ance re­quire­ment across Europe and India sim­ul­tan­eously.

Delivery

The de­liv­ery story is the sober­ing one. Gart­ner's fore­cast that more than 40 per cent of agen­tic-AI pro­jects are can­celled by end-2027, and that only ~130 of thou­sands of ‘agen­tic’ vendors are real (Gart­ner, 2025), is a warn­ing that cap­ab­il­ity is run­ning ahead of the abil­ity to de­ploy it safely. The gap between what the ar­chi­tec­ture per­mits and what or­gan­isa­tions can ac­tu­ally op­er­ate is where 2027's dis­ap­point­ments will cluster.

The four registers, reconciled

Tech­no­logy sup­plies the prim­it­ives, mar­kets sup­ply the cap­ital, reg­u­la­tion sup­plies the dead­line, and de­liv­ery sup­plies the con­straint. The or­gan­isa­tions that suc­ceed in 2027 will be those that treat all four as a single pro­gramme rather than four sep­ar­ate ones.

Three scenarios for 2027

The base case is that the architecture arrives on schedule but unevenly governed; the bull and bear cases turn on a single variable: whether governance keeps pace with capability.

The three scen­arios below are not equally likely. The base case is the work­ing as­sump­tion of this edi­tion; the bull and bear are the plaus­ible de­par­tures from it, each driven by the same ten­sion between the speed of as­sembly and the speed of gov­ernance.

Bull case: Governance catches up

The stand­ards cal­en­dar holds, and gov­ernance moves with it: the EU wal­let is is­sued on time and widely ac­cep­ted; agen­tic iden­tity be­comes a genu­ine pro­cure­ment norm; guard­ian-agent spend in­flects as fore­cast; and proven­ance reaches de­fault at a major plat­form. Trust spend­ing is re­framed as re­turn, budgets rise, and the ar­chi­tec­ture-versus-pos­ture gap be­comes a dur­able com­pet­it­ive ad­vant­age for early movers. The fraud curve bends as built-in veri­fic­a­tion blunts the gen­er­at­ive at­tack. Prob­ab­il­ity: lower.

Base case: The architecture arrives, ungoverned

The trust ar­chi­tec­ture is as­sembled on roughly the pub­lished timetable (wal­lets is­sued, agent pro­to­cols shipped, DPDP en­forced), but gov­ernance lags. More than 40 per cent of agen­tic pro­jects are can­celled (Gart­ner, 2025); wal­let ac­cept­ance is leg­ally due but un­evenly de­livered; proven­ance ad­vances without be­com­ing de­fault every­where. Fraud losses track to­ward De­loitte's US$40bn (2027). The lead­ers build trust in;

Digital TrustVolge Quarterly39 •

the lag­gards keep bolt­ing it on; the gap widens. Prob­ab­il­ity: highest.

Bear case: Capability outruns control

The ar­chi­tec­ture is as­sembled faster than any­one can gov­ern it, and the un­gov­erned edges are ex­ploited. Agent cre­den­tials pro­lif­er­ate without over­sight (90 per cent of or­gan­isa­tions re­port pres­sure to loosen iden­tity con­trols for AI and 59 per cent see no al­tern­at­ive to stand­ing priv­ileged ac­cess for ma­chine and agent iden­tit­ies, De­linea, March 2026); a high-pro­file agen­tic breach or a deep­fake-driven fraud event forces re­act­ive reg­u­la­tion; wal­let rol­louts slip and ac­cept­ance is de­ferred; the fraud curve over­shoots the US$40bn fore­cast. Trust re­verts to a de­fens­ive pos­ture under pres­sure, and the year is lost to re­medi­ation rather than ar­chi­tec­ture. Prob­ab­il­ity: lower, but rising with agent ad­op­tion.

The pivot variable

All three scen­arios share the same as­sembly of tech­no­logy and the same reg­u­lat­ory cal­en­dar. What sep­ar­ates them is gov­ernance ve­lo­city: whether over­sight of agents, wal­lets and proven­ance keeps pace with their de­ploy­ment. That single vari­able is the one an op­er­ator can ac­tu­ally in­flu­ence, which is why Part V is built around it.

What to watch, and the record

What the year turns on, the signals worth tracking as it runs, and the reference annex: sources, standards and glossary behind every figure in this edition.

The watchlist

Seven signals will tell, before the Scorecard does, which scenario 2027 is running; each is observable within the year and seeds next year's Calls.

A Call re­solves at year-end; a sig­nal moves in real time. The watch­list below is the set of lead­ing in­dic­at­ors Volge Quarterly will track through 2027 to see which way the year is bend­ing.

  • EU wal­let cer­ti­fied avail­ab­il­ity. The count of mem­ber states with a cer­ti­fied wal­let ac­tu­ally in cit­izens' hands by mid-2027: the sup­ply-side pre­con­di­tion for the ac­cept­ance duty to bite.
  • Agen­tic-iden­tity pro­cure­ment lan­guage. The first reg­u­lated tender to name OAuth-for-agents or MCP En­ter­prise-Man­aged Au­thor­iz­a­tion (spe­cific­a­tion 2026-07-28) as a re­quire­ment, con­vert­ing Call 2 from thesis to fact.
  • A named agen­tic breach. The first pub­licly at­trib­uted se­cur­ity in­cid­ent caused by a rogue or over-priv­ileged AI agent: the bear-case trig­ger.
  • Pass­key de­fault an­nounce­ments. Any top-tier con­sumer plat­form mak­ing pass­word­less the de­fault rather than an op­tion (Call 7).
  • C2PA at a plat­form de­fault. A major plat­form or CDN switch­ing Con­tent Cre­den­tials from opt-in to on-by-de­fault (Call 6).
  • DPDP rules and en­force­ment pos­ture. The final DPDP rules and the first Sig­ni­fic­ant Data Fi­du­ciary des­ig­na­tions, which set the size of the In­dian com­pli­ance mar­ket (Call 10).
  • Gulf agent pi­lots. Any Na­fath or UAE Pass ex­ten­sion into en­ter­prise or agent au­thor­isa­tion (Call 9).
A Call resolves at year-end; a signal moves in real time.

The watch­list is also the seed­bed for the 2028 edi­tion. Sig­nals that fire be­come next year's Calls; sig­nals that stay quiet be­come next year's ques­tions. This is how Volge Quarterly com­pounds: each year's ob­ser­va­tions be­come the next year's falsifi­able claims, and the Score­card keeps the whole en­ter­prise hon­est.

Sources: one of two

Every figure in this edition is sourced and year-stamped. The full source list follows, numbered for reference.

  • 1. MarketsandMarkets. Di­gital Iden­tity Solu­tions, Iden­tity Veri­fic­a­tion, IAM, Fraud De­tec­tion and Cy­ber­se­cur­ity mar­ket re­ports, 2025. https://www.mar­ket­sand­mar­kets.com/Mar­ket-Re­ports/di­gital-iden­tity-solu­tions-mar­ket-247527694.html
  • 2. Grand View Re­search. Di­gital Iden­tity Solu­tions Mar­ket and Zero Trust Se­cur­ity Mar­ket, 2025–26. https://www.grand­viewre­search.com/in­dustry-ana­lysis/di­gital-iden­tity-solu­tions-mar­ket-re­port
  • 3. Mordor In­tel­li­gence. Multi-Factor Au­then­tic­a­tion Mar­ket and Middle East Cy­ber­se­cur­ity Mar­ket, 2025. https://www.mor­dor­in­tel­li­gence.com/in­dustry-re­ports/multi-factor-au­then­tic­a­tion-mar­ket
  • 4. Mar­ket.us. Pass­keys / FIDO Au­then­tic­a­tion Mar­ket and Deep­fake De­tec­tion Mar­ket, 2025. https://mar­ket.us/re­port/deep­fake-ai-mar­ket/
  • 5. Gart­ner. In­form­a­tion Se­cur­ity and Risk Man­age­ment Spend­ing Fore­cast; Agen­tic AI pre­dic­tions, 2025. https://www.gart­ner.com/en/news­room
Digital TrustVolge Quarterly40 •
  • 6. De­loitte. Cen­ter for Fin­an­cial Ser­vices, ‘Gen­er­at­ive AI is ex­pec­ted to mag­nify the risk of deep­fakes and other fraud’, May 2024. https://www2.de­loitte.com/us/en/in­sights/in­dustry/fin­an­cial-ser­vices/deep­fake-bank­ing-fraud-risk-on-the-rise.html
  • 7. Sum­sub. Iden­tity Fraud Re­port, 2023–24. https://sum­sub.com/fraud-re­port/
  • 8. En­trust / On­fido. Iden­tity Fraud Re­port, 2025. https://www.en­trust.com/re­sources/re­ports/iden­tity-fraud-re­port
  • 9. Cy­ber­se­cur­ity Ven­tures. Of­fi­cial Cy­ber­crime Re­port: cost of cy­ber­crime, 2025. https://cy­ber­se­cur­ityven­tures.com/cy­ber­crime-dam­ages-6-tril­lion-by-2021/
  • 10. Nas­daq Ver­afin. Global Fin­an­cial Crime Re­port, 2025. https://www.nas­daq.com/global-fin­an­cial-crime-re­port
  • 11. Fed­eral Trade Com­mis­sion. Con­sumer Sen­tinel Net­work Data Book, 2024. https://www.ftc.gov/news-events/data-visu­al­iz­a­tions/data-spot­light
  • 12. Javelin Strategy & Re­search. Iden­tity Fraud Study, 2025. https://javelin­strategy.com/re­search
  • 13. European Union. Reg­u­la­tion (EU) 2024/1183 (eIDAS 2.0 / EUDI Wal­let), in force 20 May 2024. https://eur-lex.europa.eu/eli/reg/2024/1183/oj
  • 14. European Union. Reg­u­la­tion (EU) 2024/1689 (AI Act): Art­icle 50 and high-risk pro­vi­sions. https://eur-lex.europa.eu/eli/reg/2024/1689/oj
  • 15. W3C. Veri­fi­able Cre­den­tials Data Model 2.0, Re­com­mend­a­tion, 15 May 2025. https://www.w3.org/TR/vc-data-model-2.0/

Sources: two of two, and glossary

  • 16. ISO/IEC. 18013-5 (mDL) and 18013-7 (mDL over the in­ter­net), 2021 / 2024–25. https://www.iso.org/stand­ard/69084.html
  • 17. C2PA. Con­tent Proven­ance and Au­then­ti­city spe­cific­a­tion v2.x; Con­tent Au­then­ti­city Ini­ti­at­ive, 2025. https://c2pa.org/spe­cific­a­tions/
  • 18. NIST. SP 800-207 and SP 1800-35 (Zero Trust); FIPS 203/204/205 (post-quantum), 2020–25. https://csrc.nist.gov/pub­lic­a­tions/sp
  • 19. Palo Alto Net­works (CyberArk until Feb­ru­ary 2026). Iden­tity Se­cur­ity Land­scape: 109 ma­chine iden­tit­ies per human, 2026. https://www.cy­ber­ark.com/threat-re­search-blog/
  • 20. De­linea. Ma­chine Iden­tity re­search: 46:1 ratio, 68% without AI iden­tity con­trols, 2025. https://de­linea.com/re­sources
  • 21. OpenID Found­a­tion. Iden­tity Man­age­ment for Agen­tic AI white­pa­per, 7 Oc­to­ber 2025. https://openid.net/
  • 22. Model Con­text Pro­tocol. Au­thor­iz­a­tion spe­cific­a­tion (OAuth 2.1), Novem­ber 2025 re­vi­sion. https://mod­el­con­tex­t­pro­tocol.io/spe­cific­a­tion
  • 23. OWASP. Top 10 for Agen­tic Ap­plic­a­tions v1.0 (9–10 Decem­ber 2025); AIVSS scor­ing. https://genai.owasp.org/
  • 24. UIDAI / MeitY. Aadhaar; Di­gital Per­sonal Data Pro­tec­tion Act, 2023 and rules. https://www.meity.gov.in/data-pro­tec­tion-frame­work
  • 25. FIDO Al­li­ance. Pass­keys and FIDO2 au­then­tic­a­tion stand­ards, 2025. https://fidoal­li­ance.org/pass­keys/

Glossary

  • Agen­tic iden­tity: A first-class, audit­able, re­voc­able iden­tity is­sued to an autonom­ous AI agent, dis­tinct from the human or ser­vice ac­count that de­ploys it.
  • C2PA / Con­tent Cre­den­tials: A cryp­to­graphic stand­ard at­tach­ing signed, tamper-evid­ent proven­ance and edit his­tory to a piece of di­gital con­tent.
  • eIDAS 2.0 / EUDI Wal­let: The re­formed EU elec­tronic-iden­tity reg­u­la­tion and the di­gital iden­tity wal­let it man­dates for cit­izens and re­quires re­ly­ing parties to ac­cept.
  • FIDO / pass­key: Phish­ing-res­ist­ant au­then­tic­a­tion stand­ards re­pla­cing pass­words with device-bound cryp­to­graphic cre­den­tials.
  • mDL: Mo­bile driv­ing li­cence, a veri­fi­able di­gital cre­den­tial under ISO/IEC 18013-5 and -7.
  • NHI: Non-human iden­tity, a ma­chine, ser­vice or agent cre­den­tial; now the ma­jor­ity of en­ter­prise iden­tit­ies.
  • Post-quantum cryp­to­graphy (PQC): Cryp­to­graphic al­gorithms de­signed to res­ist at­tack by quantum com­puters; stand­ard­ised by NIST as FIPS 203–205.
  • Zero-trust: An ar­chi­tec­ture that au­then­tic­ates and au­thor­ises every re­quest con­tinu­ously rather than trust­ing a net­work peri­meter.
Digital TrustVolge Quarterly41 •
Digital Trust

$40 billion: Deloitte's 2027 forecast for United States generative-AI-enabled fraud, up from $12.3 billion in 2023. The loss is no longer an operational line. It is an architectural one.

Version · VSP-OTG-005 · 2026
volgePublished
every quarter
About Volge Quarterly

Volge Quarterly is Volge's signature series, published every quarter for enterprise customers. Each edition takes one subject, sets out where it currently stands on the published record, and makes a set of dated, falsifiable judgements on where it is going, written to give a buyer something firmer than a forecast: every figure carries its source and its date, every judgement carries the date on which it resolves, and the following year's edition grades each one hit, missed or partial.

Where a figure is not published, the edition says so rather than estimating it. Where published sources disagree, both are given and the disagreement is stated rather than averaged away. A number that does not survive checking is corrected here rather than repeated.

The edition carries no recommendation of any supplier and no offer of service. Where a vendor, platform or deployment is named it is because it is part of the published record of the subject.

Terms of use

This edition may be shared freely. It may not be altered or republished without permission, and figures carry their stated basis and date. Names and marks appearing in it are the property of their respective owners.