>
Verified identity, zero-trust access and provable integrity: why boards now fund trust the way they fund uptime.
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.
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.
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.
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.
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 layer | Size (base year) | Forecast | CAGR | Source |
|---|---|---|---|---|
| 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 |
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).
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.
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).
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 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.
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.
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.
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.
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.
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.
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.
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.
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
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.
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.
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.
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.
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.
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,
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).
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 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).
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.
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.
| Instrument | Milestone | Date | Significance |
|---|---|---|---|
| eIDAS 2.0 / EUDI Wallet | Relying-party acceptance duty (regulated sectors) | Late 2027 | Reg (EU) 2024/1183, in force 20 May 2024; wallets available late 2026; 80% target by 2030 |
| EU AI Act, Article 50 | Transparency duty (incl. deepfake labelling) | 2 Aug 2026 | Nearest binding date; obliges disclosure of AI-generated content |
| EU AI Act, high-risk tier | High-risk obligations apply | 2 Dec 2027 | Adopted 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, 2023 | Significant Data Fiduciary duties enforceable | 13 May 2027 | Consent, audit and identity become a live compliance market |
| NIST PQC | Standardised algorithms in certificate chains | 2024–onward | FIPS 203/204/205 published Aug 2024; migration begins |
| C2PA / Content Credentials | Spec v2.2 → v2.5, published October 2026, adding asset formats and live media; CDN preservation | 2025–27 | v2.2 published 1 May 2025; Cloudflare preserving on ~20% of web (Feb 2025) |
| Instrument | Milestone | Date | Significance |
|---|---|---|---|
| W3C Verifiable Credentials 2.0 | Recommendation | 15 May 2025 | The data model beneath wallets and mDL |
| ISO/IEC mDL (18013-5 / -7) | Presentation over the internet | 2024–25 | US 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.
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.
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.”
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
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.
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 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.
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 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,
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.
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 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.
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.
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.
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.
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.
| Date | Instrument | Who it binds | What it requires |
|---|---|---|---|
| 31 Mar 2026 | CBUAE Notice 2025/3057 | UAE banks, finance companies, exchange houses, insurers, payment service providers | Withdrawal of text-message and electronic-mail one-time passcodes for digital banking; liability for passcode-linked fraud rests with the institution |
| 13 Apr 2026 | SAMA Counter-Fraud Fundamental Requirements | Saudi banks, finance companies, payment service providers | Transaction verification beyond a passcode; re-verification on reactivation; confirmation of phone-number ownership |
| 20 Feb 2026 | IT Amendment Rules, 2026 (India) | Significant social media and other intermediaries | Labelling and non-strippable provenance metadata on synthetically generated information; takedown within three hours |
| 1 Apr 2026 | RBI authentication directions | RBI-regulated entities | Two distinct authentication factors for digital payment transactions unless exempted |
| 31 Jul 2026 | RBI Digital Payment Security Controls Directions | Banks, small finance banks, payments banks, urban co-operative banks, NBFCs | A board-approved digital payments policy covering every digital payment product |
| 5 Nov 2026 | RBI Master Direction, non-bank payment system operators | NPCI, card networks, payment aggregators, prepaid issuers | Consolidated cyber resilience and digital payment security controls |
| 13 Nov 2026 | DPDP Rules, Rule 4 (India) | Consent managers | Registration opens |
| 21 Nov 2026 | Telecommunications (User Identification) Rules, 2026 (India) | Telecom licensees | Biometric identification before a SIM is issued, changed or disconnected; the window is extendable to 21 February 2027 |
| 13 May 2027 | DPDP Rules 3, 5 to 16, 22 and 23 (India) | All data fiduciaries, every sector | Notice, 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.
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.
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.
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.
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.
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.
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.
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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
software gap that prevents an identifier being consumed also prevents a breach being seen.
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.
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.
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.
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.
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
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'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.
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.
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 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.
| Date | Entity | Amount | Breach |
|---|---|---|---|
| 6 Feb 2026 | Bank of Maharashtra | ₹32.50 lakh | Know-your-customer directions |
| 20 Mar 2026 | Tanur Co-operative Urban Bank | ₹50,000 | Failure to upload records to the Central KYC Records Registry |
| 27 Mar 2026 | Central Bank of India | ₹63.60 lakh | Know-your-customer and basic savings bank deposit account directions |
| 8 May 2026 | YES Bank | ₹31.80 lakh | Failure to implement the Central KYC Records Registry identifier |
| Jul 2026 | Six non-banking financial companies | ₹2.70 to ₹6.20 lakh each | Know-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.
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 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.
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.
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.
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.
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 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.
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
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.
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.
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.
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.
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.
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.
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.
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.
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 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.
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.
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.
| Deployment | What was implemented | When | Published outcome |
|---|---|---|---|
| Vodafone Qatar | Remote electronic know-your-customer for SIM and electronic SIM registration, Innovatrics biometrics integrated by Zentity | Dec 2025 | None |
| Zain KSA (Yaqoot) | Visitor SIM registration and activation moved online through the Ministry of Interior's Absher platform | Not dated in source | None |
| stc Bahrain | Facial biometric SIM activation with the Telecommunications Regulatory Authority, claimed a regional first | Mar 2024 | None |
| Malaffi, Abu Dhabi | Health information exchange matching records by Emirates ID | Ongoing | 3,000+ facilities, 50,000+ clinical users, 3.5bn clinical records; cost not public |
| UAE Cyber Security Council and QuantumGate | National Crypto Discovery Tool to National Cryptography Center specification | May 2026 | Contract value not public |
| Deployment | What was implemented | When | Published outcome |
|---|---|---|---|
| Indian Digital Payment Intelligence Corporation | Section 8 non-profit applying AI to real-time payment fraud detection, building on the RBI's MuleHunter framework | Oct 2025 | Announced only; authorised capital ₹500 crore, paid-up ₹200 crore, ₹10 crore seed each from SBI and Bank of Baroda |
| Zain Jordan | Iris recognition for SIM registration with IrisGuard, claimed a world first for an operator | Aug 2026 | None; 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.
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.
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.
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.
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.
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.
| Source | Ratio | Basis |
|---|---|---|
| Palo Alto Networks, 2026 Identity Security Landscape | 109:1 | Survey of 2,930 leaders; machine identities including AI agents |
| Cloud Security Alliance, Non-Human Identity working group | 45:1 | Median enterprise ratio, against a 17:1 baseline in 2023 |
| Veza, State of Identity and Access 2026 | 17:1 | Not stated in the available summary |
| ManageEngine | Above 100:1 for nearly half of organisations | Some sectors reported at 500:1 |
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.
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,
because the assurance of the first is routinely assumed to extend to the second.
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.
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.
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 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.
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
adoption inside one system. The Gulf problem is breadth of integration across six.
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.
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 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.
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.
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.
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.
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
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 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 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 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
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 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 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
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'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.
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
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 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'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.
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'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.
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
fail in at least one other, and that the failure will not be technical.
| State | Scheme and issuer | Published users or volume | Financial-services instrument | Accepted remote routes | Published enforcement |
|---|---|---|---|---|---|
| Saudi Arabia | Nafath, National Information Center, overseen by the data and AI authority | 32 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. Unreconciled | Counter-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 2026 | Government identity route against the national register; the account of Nafath for nationals and Absher for residents rests on secondary commentary only | 48 decisions under Article 36 of the data protection law, announced 16 January 2026; amounts not disclosed |
| United Arab Emirates | UAE PASS, with the telecommunications and digital government regulator | More 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. Unreconciled | Notice 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 2022 | Digital identity systems permitted, including UAE PASS and the federal identity authority's validation gateway; Emirates identity document and verification record retained | 200 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 |
| Qatar | Tawtheeq, the National Authentication System, transport and communications ministry; Qatar Digital ID, Ministry of Interior | Not published | Electronic know-your-customer procedures issued 15 October 2023; Payment Services Regulation requires compliance with the eKYC security principles | Ministry of Interior identity card for citizens and residents, commercial registration for companies, passport for visitors and non-residents | Not published |
| State | Scheme and issuer | Published users or volume | Financial-services instrument | Accepted remote routes | Published enforcement |
|---|---|---|---|---|---|
| Bahrain | eKey, the National Authentication Framework, Information and eGovernment Authority | Not published | Financial 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 2025 | National 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 geolocation | Not published |
| Oman | THEQA, national computer emergency readiness team, transport and communications ministry | 202.9 million authentication requests to end 2025, attributed to the national digital transformation report. No user count published | Instructions on digital onboarding and electronic know-your-customer, June 2023 | Verification against Mala'a, the national digital onboarding registry, under central bank supervision; business onboarding at proof of concept with three members | Not published |
| Kuwait | Kuwait Mobile ID, Public Authority for Civil Information | Not published | Central bank directive to exchange companies, reported 22 and 23 April 2026; no published compliance date | Authentication through the government mobile identity application before each online international transfer, with direct electronic integration to the civil information authority | Not 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.
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
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 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
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.
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.
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 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.
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.
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
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.
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 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.
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.
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.
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
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.
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.
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.
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.
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.
Each Call states a claim, a confidence band and the reasoning that supports it. The claims are written to be graded, not admired.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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 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.
| Band | Meaning | Residual uncertainty |
|---|---|---|
| High | Evidence base is measured and corroborated | Institutional behaviour, not direction |
| Medium–High | Direction is clear and dated | Timing, phasing, or a demanding second clause |
| Medium | Direction is likely | A high threshold (‘default’, ‘standard’) or an unproven step |
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
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 same year read from four angles (technology, markets, regulation and delivery) and stress-tested against three scenarios for how 2027 could run.
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.
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.
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.
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.
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.
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.
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.
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.
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;
the laggards keep bolting it on; the gap widens. Probability: highest.
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.
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 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.
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.
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.
Every figure in this edition is sourced and year-stamped. The full source list follows, numbered for reference.
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.
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.