No single regulator owns AI in banking. Banks answer to a layered system: binding AI law where it exists (the EU AI Act, whose high-risk regime — including credit scoring — now applies from December 2, 2027 after the 2026 Digital Omnibus deferral, Colorado's ADMT Act from January 1, 2027, and California's ADMT regulations from the same day), prudential supervisors applying existing frameworks to AI (the Federal Reserve, OCC, FDIC, NCUA, ECB, and UK PRA/FCA), consumer-protection, market and financial-crime authorities that regulate outcomes regardless of how a decision was made (CFPB, SEC, CFTC, FinCEN, NY DFS), and global standard-setters shaping what supervisors expect next (FSB, Basel Committee). This page tracks all 60 of them and every document they have published.
Which AI consultations are open for comment?
- ISA Draft Directive on Technological Means and AI — Draft Directive to Licensees on Providing Services Using Technological Means and Artificial Intelligence (New Version) · closes Dec 6, 2026
- HKMA discussion paper on banks' use of Agentic AI — Discussion Paper on Agentic A.I. Governance in the Banking Sector · closes Feb 1, 2027
What are the next AI regulation deadlines for banks?
| Date | Authority | Deadline |
|---|---|---|
| Dec 2026 | Bank of Thailand | Deadline for D-SIBs and significant e-money providers to run technology-based abnormal-transaction detection · Milestone |
| Dec 1, 2026 | Colorado AI Act | Regulation 10-1-1: annual life insurer reports and non-use attestations due · Milestone |
| Dec 2, 2026 | EU AI Act | Article 50(2) marking transition ends · Milestone |
| Dec 6, 2026 | Israel (BoI / ISA / CMISA) | ISA Draft Directive on Technological Means and AI: comment period closes · Comment deadline |
| Jan 2027 | Central Bank of Ireland | Applications open for the Advancing AI for Consumer Good sandbox · Milestone |
| Jan 1, 2027 | Colorado AI Act | SB 26-189 takes effect · Takes effect |
Which US authorities regulate AI in banking?
| Authority | Jurisdiction | Force | Key document | Latest move |
|---|---|---|---|---|
| Federal Reserve | United States (state member banks, bank holding companies, large financial institutions) | Supervisory guidance | SR 26-2 — Revised interagency Model Risk Management guidance (Apr 2026, supersedes SR 11-7) | Nov 13, 2026: the Board's Financial Stability Report found AI the risk market contacts cited most often and said reports of agentic systems involved in cyber events have increased since May; Nov 16, 2026: Governor Barr said major banks report AI agents attempting to take actions in retail customer accounts without the bank being able to verify authenticity, consent and alignment with customer instructions; Sep 29, 2026: Governor Waller's Sibos speech on AI payment agents distinguishes 'agent-assisted' from 'agent-delegated' commerce and flags authentication, liability and fraud-detection as open trust problems, calling for an industry-led, tested (not mandated) approach; same day, Vice Chair for Supervision Bowman at the Community Bank Cyber Workshop: AI used by threat actors 'adds complexity to the risk environment' and is becoming 'both a defensive tool and an evolving risk'; Apr 2026 revised model risk guidance |
| OCC | United States (national banks and federal savings associations) | Supervisory guidance | OCC Bulletin 2026-54 — Super Intelligence: Supervisory Framework for examiners (Nov 2026); Bulletin 2026-13 for model risk (Apr 2026) | Nov 4, 2026: Bulletin 2026-54, the Super Intelligence supervisory framework for examiners on generative and agentic AI; Apr 2026 revised model risk guidance excluding those systems |
| FDIC | United States (state-chartered banks that are not Federal Reserve members; deposit insurer for all insured banks) | Supervisory guidance | FIL-15-2026 — Agencies Revise the Interagency Model Risk Management Guidance (Apr 17, 2026) | Apr 2026 adoption of revised interagency model risk guidance (FIL-15-2026); Jun 2026 testimony describing it as 'an avenue for the safe and sound adoption of technology' |
| NCUA | United States (federally insured credit unions) | Supervisory guidance | Credit Union Super Intelligence Resource Center, formerly the Credit Union Artificial Intelligence Resource Center (Aug 2025, renamed Nov 18, 2026): https://ncua.gov/regulation-supervision/regulatory-compliance-resources/super-intelligence-si — the NCUA's only AI-specific publication for credit unions | Nov 18, 2026: AI pages renamed 'Super Intelligence (SI)' (ncua.gov/ai now redirects to ncua.gov/si), guidance otherwise unchanged in substance; Feb 2026 Senate testimony frames 'creating space for credit unions to innovate responsibly' with AI as a 2026-2030 strategic-plan goal; Apr 2026 refresh of the AI resource page; no AI mention in the Jan 2026 supervisory priorities |
| CFPB | United States (consumer financial products) | Binding law | Regulation B §1002.9 adverse-action requirements (ECOA); April 2026 Reg B final rule | April 2026 Regulation B final rule (effective July 21, 2026) eliminates disparate-impact liability under ECOA — the fair-lending theory most often applied to AI models — while the statutory duty to give specific, accurate adverse-action reasons remains untouched. |
| SEC | United States (broker-dealers, investment advisers, funds, and public companies — including bank holding companies and bank-affiliated securities arms) | Supervisory guidance | Division of Examinations FY2027 Examination Priorities (Oct 23, 2026) — the operative statement of what SEC examiners test, now framed around 'super intelligence (SI)' | Nov 9, 2026: closing a Commission roundtable on 'Machine-Understandable Data in the Age of Super Intelligence (SI)' (also featuring Commissioner Uyeda and Chief Economist Joshua White), Chairman Atkins said SI 'offers extraordinary potential' but warned the same technology 'can also magnify errors and missing context' in disclosure data — discussion only, no new rule or disclosure requirement; Oct 23, 2026 FY2027 Division of Examinations priorities adopted 'super intelligence (SI)' as the Division's term, with examinations focused on SI-related cyberattacks, deepfake social engineering, and monitoring of SI in fraud, AML, Reg BI and trading; two more AI-related enforcement actions landed in the same window (GenesisAI/Cheishvili settlement, Aug 26, 2026; Cryptoaiml/TSAI Pro fraud charges, Sep 29, 2026) |
| CFTC | United States (futures, options, and swaps markets; bank swap dealers, FCMs, exchanges, and clearinghouses) | Supervisory guidance | CFTC Staff Advisory on the Use of AI in CFTC-Regulated Markets (CFTC Letter No. 24-17, Dec 2024) | Nov 4, 2026: NYMEX consented to a further CFTC review extension, to January 8, 2027, for the two compute futures contracts it submitted for Commission approval on August 11, 2026 (Silicon Data H100 and B200 Rental Index Futures, NYMEX Submission No. 26-370), still under review; follows Chairman Selig's Oct 28, 2026 Frontier Forum keynote on 'super intelligence' and agentic finance, which raised whether a 'know-your-agent' (KYA) standard is needed and who is liable when an autonomous trading agent malfunctions |
| FinCEN | United States (Bank Secrecy Act administrator for all US financial institutions) | Binding law | AML/CFT Program proposed rule (Apr 2026) — names effective use of AI as a factor in FinCEN enforcement decisions; supersedes the withdrawn July 2024 proposal | Sep 3, 2026 alert FIN-2026-Alert005 names AI-supported software platforms as part of the ecosystem enabling digital-asset investment scam centers, and AI-tool adoption as fueling their growth; Oct 1, 2026 alert FIN-2026-Alert007 on the Russia-linked A7 Network lists AI-altered invoices among its red flags; the Apr 2026 AML/CFT program NPRM remains unfinalized, with industry trackers reporting a final rule is not expected before 2027 |
| U.S. Treasury | United States (federal; convenes FSOC and the financial-sector critical-infrastructure partnership; parent of the OCC and FinCEN) | Voluntary framework | Financial Services AI Risk Management Framework (FS AI RMF) and AI Lexicon — voluntary, NIST-aligned resources released Feb 19, 2026 | June 2026: FSOC and Treasury's AI Transformation Office concluded the four-roundtable AI Innovation Series (Mar–May 2026); participants asked for regulatory clarity and harmonization to scale AI adoption |
| White House | United States (federal executive branch; directs executive departments and agencies, not banks) | Non-binding standards | EO 14365, 'Ensuring a National Policy Framework for Artificial Intelligence' (signed December 11, 2025; 90 FR 58499), with EO 14179 (90 FR 8741) and America's AI Action Plan (July 23, 2025); EO 14434 (signed September 29, 2026; 91 FR 63129) sets the 'Super Intelligence' terminology | On September 29, 2026 the President signed EO 14434, 'Inaugurating the Era of Super Intelligence' (91 FR 63129), directing executive-branch agencies to use the terms 'Super Intelligence' and 'SI' in place of 'Artificial Intelligence' and 'AI' in official correspondence, websites, reports and other non-statutory documents, with 'SI' defined by reference to the existing statutory definition of AI (15 U.S.C. 9401(3)); the Assistant to the President for Science and Technology must submit proposed legislative language for a Federal definition within 60 days |
| FINRA | United States (all registered broker-dealer member firms; relevant to banks through bank-affiliated broker-dealers and securities subsidiaries) | Supervisory guidance | Regulatory Notice 24-09, 'FINRA Reminds Members of Regulatory Obligations When Using Generative Artificial Intelligence and Large Language Models' (June 27, 2024), and the GenAI section of the 2026 FINRA Annual Regulatory Oversight Report (December 9, 2025) | On July 9, 2026 FINRA published Regulatory Notice 26-14 proposing to modernize Rule 2210 (Communications with the Public), citing advances in generative AI and replacing the prescriptive principal pre-use approval of retail communications with risk-based supervision standards; comments were due September 11, 2026 |
| NY DFS | New York State (state-chartered banks, foreign bank branches and agencies, insurers, money transmitters, virtual-currency licensees, and other DFS-licensed entities) | Supervisory guidance | 23 NYCRR Part 500 (Second Amendment effective Nov 1, 2023; fully phased in Nov 1, 2025) as applied to AI by the Oct 16, 2024 Industry Letter | Nov 10, 2026: DIGIT launches the RAISE Portal, requiring large frontier AI developers to register by January 1, 2027, and Governor Hochul announces the New York Center for AI Responsibility and Research (NYCAIRR) at Binghamton University to independently test frontier AI systems; Sep 21, 2026: Governor Hochul announced RAISE Act implementation next steps and named Marc Gilman Deputy Director of DFS's new DIGIT office |
| NYC DCWP | New York City, United States (employers and employment agencies using automated employment decision tools to screen candidates for employment or employees for promotion within the city) | Binding law | Local Law 144 of 2021 (NYC Administrative Code §§ 20-870 to 20-874) and DCWP's final rule (6 RCNY §§ 5-300 to 5-304); enforcement began July 5, 2023 | The New York State Comptroller issued Report 2024-N-6 on December 2, 2025, finding DCWP's system for enforcing Local Law 144 ineffective and recommending enforcement that does not rely only on complaints |
| Colorado AI Act | Colorado, United States (any developer or deployer doing business in Colorado) | Binding law | SB 26-189, the Automated Decision-Making Technology Act (signed May 14, 2026; effective Jan 1, 2027) | Oct 26, 2026: the Colorado Department of Law held its rulemaking hearing on the ADMT and Chatbot Safety Act rules; lenders, healthcare providers and business groups testified that the draft would burden national companies, while the Colorado Technology Association framed the debate as consumer protection versus innovation. No final rule had been adopted as of Nov 2, 2026, ahead of the Jan 1, 2027 statutory deadline |
| State of California | California, United States (any business meeting CCPA thresholds that handles Californians' personal information; employers with five or more California employees) | Binding law | CPPA regulations on cybersecurity audits, risk assessments and automated decisionmaking technology (11 CCR §§ 7001, 7120–7124, 7150–7157, 7200–7222; approved Sep 22, 2025, effective Jan 1, 2026; ADMT obligations from Jan 1, 2027) | Governor Newsom signed AB 1609 (customer-service chatbot disclosure and human hand-off, businesses over $500M revenue) on September 28, 2026 (Chapter 733) and SB 947 (the 'No Robo Bosses Act' — bars sole reliance on automated decision systems for employee discipline or termination) on September 30, 2026 (Chapter 859); AB 1018 did not pass the Legislature. ADMT obligations for significant decisions apply from Jan 1, 2027 |
| Texas AG | Texas, United States (any person who promotes, advertises or conducts business in Texas, produces a product or service used by Texas residents, or develops or deploys an AI system in Texas) | Binding law | Texas Responsible Artificial Intelligence Governance Act (HB 149, 89th Legislature; Business & Commerce Code Title 11, Subtitle D, Chapters 551–554; signed June 22, 2025, effective January 1, 2026) | September 1, 2026 was the statutory deadline (HB 149, Section 8) for the Attorney General to post the information and online complaint mechanism required by Section 552.102; the Attorney General's consumer-protection site now carries a TRAIGA overview and an online AI complaint form. As of October 5, 2026 this review found no published enforcement action under the Act |
| Utah Division of Consumer Protection | Utah, United States (suppliers in consumer transactions with Utah consumers, and occupations licensed by the Department of Commerce) | Binding law | Utah Artificial Intelligence Policy Act (SB 149, 2024; Title 13, Chapter 72) and the generative-AI consumer-disclosure chapter (SB 226, 2025; Title 13, Chapter 77, effective May 7, 2025) | Sept 16, 2026: the Office of Artificial Intelligence Policy delivered its annual report to the Business and Labor Interim Committee ahead of the Nov 30 deadline, and the committee voted 15-0 to recommend extending the Artificial Intelligence Policy Act's sunset from July 1, 2027 to July 1, 2037 — a recommendation, not yet enacted; HB 320 (2026 General Session), signed March 18, 2026 and effective May 6, 2026, amended the Office's learning-laboratory provisions to add joint interpretation agreements, regular audits of participants and up to two extensions of a demonstration period; the HB 286 frontier-model transparency bill was filed without passing on March 6, 2026 |
| Illinois IDHR | Illinois, United States (employers with one or more employees in Illinois for 20 or more calendar weeks in a year, or any employer for disability, pregnancy and sexual-harassment claims; employers that ask applicants for Illinois-based positions to record video interviews) | Binding law | Illinois Human Rights Act amendments on artificial intelligence in employment (HB 3773, Public Act 103-0804; 775 ILCS 5/2-101(M), (N) and 2-102(L); approved August 9, 2024, effective January 1, 2026) | On June 2, 2026 the Department of Human Rights withdrew its May 15, 2026 proposed rules implementing the Section 2-102(L)(2) AI-notice requirement and canceled the June 10, 2026 hearing, citing the need for more collaboration with other state agencies; as of November 2, 2026 no revised proposal has been filed and the statutory notice duty applies without rule-level detail |
| Massachusetts AG | Massachusetts, United States (any business that markets, sells or lends to Massachusetts residents, wherever it is based) | Binding law | Attorney General Advisory on the Application of the Commonwealth's Consumer Protection, Civil Rights, and Data Privacy Laws to Artificial Intelligence (April 16, 2024), and the Earnest Operations LLC Assurance of Discontinuance ($2.5 million; announced July 10, 2025) | July 10, 2025: Attorney General Andrea Joy Campbell announced a $2.5 million settlement with Earnest Operations LLC over AI underwriting models, with mandated AI governance, annual fair-lending testing and reporting to the Attorney General |
| New Jersey DCR | New Jersey, United States (employers, housing providers, places of public accommodation, credit providers, contractors and other entities covered by the Law Against Discrimination) | Binding law | Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination (January 2025; announced January 9, 2025), and the Division's disparate-impact rules, N.J.A.C. 13:16 (adopted November 5, 2025; effective December 15, 2025) | December 15, 2025: the Division's disparate-impact rules (N.J.A.C. 13:16) took effect, following adoption on November 5, 2025; they build on the January 2025 algorithmic-discrimination guidance |
| NIST | United States (voluntary, used globally) | Voluntary framework | AI Risk Management Framework 1.0 (Jan 2023) + Generative AI Profile (Jul 2024) | Growing role as the default governance scaffold after the 2026 US model-risk revision excluded generative AI |
Which international bodies set AI rules for banks?
| Authority | Jurisdiction | Force | Key document | Latest move |
|---|---|---|---|---|
| EU AI Act | European Union | Binding law | Regulation (EU) 2024/1689 (in force Aug 1, 2024), as amended by the Digital Omnibus on AI (EU) 2026/1744 | Regulation (EU) 2026/1744 (Digital Omnibus on AI) entered into force July 27, 2026, deferring Annex III high-risk obligations — including credit scoring — from Aug 2, 2026 to Dec 2, 2027 |
| European Union | European Union (all 27 member states; regulations apply directly, directives through national transposition) | Binding law | Consumer Credit Directive (EU) 2023/2225, Article 18 (applies from Nov 20, 2026) — the first EU rule written for automated creditworthiness assessment outside the AI Act | Nov 20, 2026: the recast Consumer Credit Directive's rules, including the right to human intervention and an explanation for automated creditworthiness assessments, apply in member states that transposed it by the November 20, 2025 deadline |
| ECB | Euro area (significant institutions) | Supervisory guidance | SSM supervisory priorities 2026–28 (AI under the operational-resilience priority) | 31 October 2026: the deadline for significant institutions to submit AI-cyber action plans under the 7 July 2026 'Dear CEO' letter (SSM-2026-0301) passed; Supervisory Board member Pedro Machado said on 30 October 2026 that 'a more meaningful and substantial assessment can only be made after we get the banks' action plans,' so the ECB's horizontal analysis is still pending |
| EBA | European Union | Supervisory guidance | Factsheet on the AI Act's implications for banking and payments (Nov 2025) | Nov 19, 2026: the Autumn 2026 Risk Assessment Report found nearly three quarters of surveyed banks have access to frontier AI models and around 80% source AI models from the US, and warned that delegating tasks to AI agents risks loss of human control; Nov 5, 2026: the revised internal governance guidelines (EBA/GL/2026/10) bring risks from the use of AI and quantum technologies into the scope of banks' governance frameworks; Jul 31, 2026: joint ESA statement (JC 2026 25) on ICT risks from frontier AI models — banks told to adjust DORA ICT-risk controls around prevention, detection and management without delay |
| ESMA | European Union (investment firms and credit institutions providing investment services under MiFID II; markets and investor-protection supervisory convergence) | Supervisory guidance | ESMA Public Statement on the use of Artificial Intelligence in the provision of retail investment services (30 May 2024, ESMA35-335435667-5924) | 23 September 2026: ESMA announced a new Union Strategic Supervisory Priority on digital innovation from 2027, with an initial focus on how supervised entities use AI and tokenisation |
| EIOPA | European Union | Supervisory guidance | Opinion on AI governance and risk management (EIOPA-BoS-25-360, 6 August 2025) | 6 August 2025 — EIOPA published its Opinion on AI governance and risk management (EIOPA-BoS-25-360) for national insurance supervisors; on 31 July 2026 EIOPA joined the EBA and ESMA in the ESAs' joint statement on ICT risks from frontier AI models (JC 2026 25). |
| UK (BoE / PRA / FCA) | United Kingdom | Supervisory guidance | BoE/PRA response to the Treasury Select Committee on AI (Apr 2026) | November 19, 2026: the PRA, FCA and Bank of England published non-binding effective practices on cyber risk and resilience relating to AI (continuous exposure management at machine speed, governed AI-assisted defence, AI terms in third-party contracts, meaningful human oversight) alongside the joint cyber consultation CP17/26 / GC26/9, which closes on February 26, 2027; October 2026: the FCA published illustrative AI good- and poor-practice examples (Oct 19) that set out governance expectations for agentic AI — a central agent inventory and governance designed before deployment, not after — and the BoE/FCA published the fourth AI survey's results (Oct 26): 86% of 196 respondent firms use AI, 80% have AI agents deployed (plus 14% planning), but only 2% of those agents have 'broad autonomy'; cyber-risk amplification and third-party concentration are now the top two systemic risks; September 30, 2026: Governor Bailey's 'Frontier AI and the Question of Governance' Bank Insights article says 'regulation is not, in my view, the right place to start' and ties frontier AI directly to financial stability through cyber risk and agentic trading and payments |
| BaFin | Germany (banks, insurers and other financial entities supervised by BaFin or, for significant banks, the ECB) | Supervisory guidance | Guidance on ICT Risks in the Use of AI at Financial Entities (18 December 2025; English translation published 30 January 2026), alongside the 2021 BDAI principles paper and the 2022 ML risk-models results paper | 29 July 2026: BaFin published its page on market surveillance of AI systems, stating that it monitors Article 5 prohibited practices, Article 50 transparency obligations and Article 4 AI literacy now, and high-risk AI systems from 2 December 2027 (including creditworthiness assessment of natural persons) |
| ACPR | France (banks, insurers and other supervised financial entities; significant banks are supervised jointly with the ECB) | Supervisory guidance | ACPR discussion document 'Governance of Artificial Intelligence in Finance' (June 2020) | 1 July 2026: the ACPR launched a public consultation on algorithmic fairness in the financial sector (open until 30 September 2026) and held a market meeting on AI regulation and supervision |
| DNB / AFM | Netherlands (DNB: prudential supervision of banks, insurers and pension funds; AFM: conduct and market supervision; DNB also acts within the ECB's Single Supervisory Mechanism for significant banks) | Supervisory guidance | AFM and DNB, 'The impact of AI on the financial sector and supervision' (9 April 2024), building on DNB's 'General principles for the use of Artificial Intelligence in the financial sector' (25 July 2019) | 19 January 2026: the AFM announced in its Agenda 2026 that it will intensify supervision of the responsible use of AI and of DORA compliance; April 2026: AFM report on AI in the Dutch asset management sector |
| FINMA | Switzerland (banks, securities firms, insurers, fund managers, managers of collective assets, financial market infrastructures and other supervised institutions) | Supervisory guidance | FINMA Guidance 08/2024: Governance and risk management when using artificial intelligence (18 December 2024) | 24 April 2025: FINMA published a survey of around 400 Swiss financial institutions showing about half use AI or have initial applications in development, following its Guidance 08/2024 of 18 December 2024 |
| OSFI | Canada (federally regulated financial institutions: banks, foreign bank branches, insurers, and trust and loan companies) | Supervisory guidance | Guideline E-23 – Model Risk Management (2027): published 11 September 2025, effective 1 May 2027; names AI/ML in the model definition and sets 3 outcomes and 12 principles | Oct 8, 2026 2026 Semi-Annual Risk Outlook: frontier AI is amplifying cyber, technology, third-party and reputational risk, citing autonomous AI capabilities that 'rapidly and autonomously chain together multiple vulnerabilities' and provider concentration in frontier-AI models and cloud infrastructure; in July 2026 OSFI issued a Technology Risk Bulletin on generative and agentic AI, following its April 2026 bulletin on frontier AI and the March 23, 2026 FIFAI II report; Guideline E-23 takes effect 1 May 2027 |
| MAS | Singapore (central bank, integrated financial regulator and supervisor of banks, insurers, capital markets firms and payment providers) | Supervisory guidance | Guidelines on Artificial Intelligence Risk Management (issued 7 October 2026; Sections 3-4 apply from 7 October 2027, Sections 5-6 by 7 October 2028) | 7 October 2026: MAS issued final Guidelines on Artificial Intelligence Risk Management for financial institutions, closing Consultation Paper P017-2025 with a response-to-feedback paper and a phased transition running to October 2028. |
| PDPC Singapore | Singapore (all private-sector organisations that collect, use or disclose personal data, including banks) | Binding law | Advisory Guidelines on Use of Personal Data in AI Recommendation and Decision Systems (1 March 2024), supplemented by the Advisory Guidelines on Use of Personal Data in Generative AI (20 July 2026) | 20 July 2026: the PDPC issued Advisory Guidelines on Use of Personal Data in Generative AI, allocating PDPA responsibilities between model providers, system providers and system deployers. |
| IMDA Singapore | Singapore (cross-sector digital and AI governance; not a financial regulator) | Voluntary framework | Model AI Governance Framework for Agentic AI (launched 22 January 2026; version 1.5 published 20 May 2026) | October 2026: IMDA and the AI Verify Foundation published the Starter Kit for Testing Agentic Systems for Safety and Reliability, alongside the AI Verify Foundation's first five accredited third-party AI testers (AI TAP) and a new Frontier AI Resilience and Security Taskforce (FAIRS) — direct follow-through on the agentic-AI testing guidelines IMDA flagged when it launched the Model AI Governance Framework for Agentic AI in January 2026. |
| HKMA | Hong Kong SAR (all Authorized Institutions: licensed banks, restricted licence banks and deposit-taking companies) | Supervisory guidance | "ART" Framework and updated Supervisory Guidance for Responsible Adoption of Artificial Intelligence (15 October 2026), building on the 2019 "High-level Principles on Artificial Intelligence" and the 19 August 2024 generative-AI extension | On 2 November 2026, during Hong Kong FinTech Week, the HKMA launched Project SAGE (Securing Agentic Finance with Governance and Enablement) and a cross-sectoral Agentic Finance Expert Group, and published a discussion paper on banks' use of agentic AI proposing an 'intent-anchored accountability' model, a registry of material AI agents, kill-switch intervention mechanisms and 'Know Your Agent' principles — the follow-through flagged alongside the 15 October 2026 ART Framework; comments are due 1 February 2027 |
| Hong Kong SFC | Hong Kong SAR (regulator of licensed corporations, asset managers and SFC-licensed virtual asset trading platforms; banks are supervised by the HKMA) | Supervisory guidance | SFC Circular 24EC55, 'Use of generative AI language models' (12 November 2024) | 2 June 2026: SFC Circular 26EC32 on enhanced cybersecurity measures against AI-enabled cyberattacks, addressed to licensed corporations, SFC-licensed VATPs and associated entities. |
| Japan FSA | Japan (supervisor of banks, securities firms and insurers) | Supervisory guidance | AI Discussion Paper (Version 1.1), 3 March 2026, updating Version 1.0 of 4 March 2025 | 22 May 2026: the FSA and the Bank of Japan jointly requested financial institutions to implement nine short-term responses to frontier AI cyber threats, expected over roughly one month; the Bank of Japan's October 2026 Financial System Report reports institutions assessing internet-facing systems against the nine-measure checklist but gives no completion data, with on-site and off-site monitoring continuing. |
| Taiwan (FSC / MODA) | Taiwan (FSC: banks, securities and futures firms and insurers; MODA and NSTC: cross-sector AI governance under the AI Basic Act) | Supervisory guidance | Guidelines for AI Applications in the Financial Industry (FSC, 20 June 2024), read with the Artificial Intelligence Basic Act (promulgated 14 January 2026) | 13 Oct 2026: the FSC published a consultation document on the security of agentic AI in the financial industry (use-case governance, a four-level L1 to L4 risk classification on three axes, zero-trust controls), with comments taken for 30 days to 12 November 2026; 14 Jul 2026: FSC recommendations on frontier-AI attack risk; 7 Jul 2026: the Ministry of Digital Affairs issued the AI Risk Taxonomy and Assessment Framework under Article 16 of the AI Basic Act. |
| APRA / ASIC | Australia (APRA: authorised deposit-taking institutions, insurers and superannuation trustees; ASIC: financial services and credit licensees) | Supervisory guidance | APRA Letter to Industry on Artificial Intelligence (AI), 30 April 2026, with the binding standards CPS 230 (Operational Risk Management) and CPS 234 (Information Security) and ASIC Report 798 (29 October 2024) | Nov 12, 2026: APRA's System Risk Outlook names AI and reliance on material service providers as changing the nature and scale of operational risk, citing the June 2026 OpenAI agent incident that accessed an Australian Government website as evidence agents can cause unintended harm without effective human oversight, and says APRA 'is translating this assessment into a more intensive and forward-looking supervisory approach to AI-related risks'; on 27 August 2026 APRA and ASIC urged financial entities to move from awareness of frontier AI risks to decisive action after nine roundtables in June and July 2026; on 30 April 2026 APRA's AI letter called for a step-change in AI risk management and the amended CPS 230 took effect on 1 July 2026 |
| RBI | India (commercial banks including foreign banks, small finance banks, payments banks, local area banks, regional rural banks, co-operative banks, all-India financial institutions, NBFCs, asset reconstruction companies and credit information companies) | Voluntary framework | FREE-AI Committee report "Framework for Responsible and Ethical Enablement of Artificial Intelligence (FREE-AI) in the Financial Sector" (13 August 2025), and the draft Guidance on Regulatory Principles for Model Risk Management, 2026 (24 June 2026) | On 24 June 2026 the RBI released the draft Guidance on Regulatory Principles for Model Risk Management, 2026, covering AI/ML and third-party models, with comments invited until 24 July 2026; no final version had been published on the RBI website as of 5 October 2026 |
| FSB | Global (G20) | Non-binding standards | Sound Practices for Responsible Adoption of Artificial Intelligence (AI): Final report (Oct 15, 2026) | 17 November 2026: after its 16-17 November Plenary in Basel, the FSB said advanced AI models have increased risks to the financial system, discussed possible further work on operational resilience, the nexus between AI and fraud and whether existing FSB work should be updated for the latest AI developments, and named AI among its 2027 priorities (work programme due in early 2027); 15 October 2026: the FSB published the final report finalising its 'Sound Practices for the Responsible Adoption of AI,' with a companion overview of the 124 consultation responses published the same day — a 2026 G20 deliverable confirmed finalised in the FSB Chair's 12 October 2026 letter to the G20 |
| Basel Committee | Global (28 jurisdictions) | Non-binding standards | Digitalisation of finance report (May 2024) | Nov 18, 2026: at its Basel meeting the Committee said information security and third-party dependency remain leading supervisory priorities, that the expanding use of digital channels, cloud services and AI can increase exposure to cyber attacks, digital fraud and correlated dependencies, and discussed banks' links to financing of data centres and AI; Oct 19, 2026: AML/CFT supervisory newsletter (No 37) on AI in transaction monitoring; Oct 1, 2026: following its 28–29 September meeting, the Committee agreed to review the operational risk framework's 'event type' loss categories with a focus on cyber risk and AI developments, its first step toward amending an existing hard standard because of AI; June 2026 ICT risk-management report |
| FATF | Global (FATF members and the FATF Global Network of regional bodies) | Non-binding standards | Horizon Scan: Artificial Intelligence and Deepfakes (December 2025), read with Recommendation 15 of the FATF Recommendations (consolidated October 2025) | Dec 22, 2025: horizon scan on AI and deepfakes published after approval at the 22–24 October 2025 Plenary; the February and June 2026 Plenary outcomes record no further AI item |
| IOSCO | Global (international standard-setter for securities regulators; members include the SEC, ESMA, the FCA and other national securities regulators) | Non-binding standards | FR/02/2026, 'Supervisory Toolkit for AI Use in Capital Markets' (Final Report, May 25, 2026), building on FR06/2021 (September 2021) and CR/01/2025 (March 2025) | On May 25, 2026 IOSCO published its final Supervisory Toolkit for AI Use in Capital Markets (FR/02/2026) with a standalone extract (OR/07/2026) for use in examinations and inspections; feedback from stakeholders was invited by June 26, 2026, and the next phase is a review of emerging industry practices |
| Central Bank of Ireland | Ireland (banks, credit unions, payment and e-money institutions, insurers, investment firms, funds and other firms regulated by the Central Bank of Ireland) | Supervisory guidance | Regulatory & Supervisory Outlook 2026, Spotlight 1 'Approaching AI from a Supervisory Perspective' (26 February 2026), read with S.I. No. 366/2025 and the Regulation of Artificial Intelligence Act 2026 | 2 Nov 2026: Governor Makhlouf told the Financial System Conference that the vast majority of regulated firms already use AI, reported supervisory engagement on frontier-AI cyber risk and announced 'Advancing AI for Consumer Good' as the theme of the third Innovation Sandbox Programme (applications open January 2027); 31 Jul 2026: the Regulation of Artificial Intelligence Act 2026 came into operation, giving the Central Bank AI Act enforcement and fining powers. |
| Luxembourg (CSSF / BCL) | Luxembourg (credit institutions, investment firms, payment and e-money institutions, investment fund managers, specialised and support PFS and other entities supervised by the CSSF) | Supervisory guidance | Thematic review on the use of Artificial Intelligence in the Luxembourg financial sector (CSSF and BCL, 16 May 2025), read with the CSSF communiqué on frontier AI of 7 July 2026 | 17 Nov 2026: the CSSF, jointly with the BCL, opened a new AI questionnaire on eDesk for banks, investment firms, payment and e-money institutions, specialised and support PFS, crypto-asset service providers and authorised fund managers, due by 31 December 2026; 7 Jul 2026: CSSF communiqué setting management-body expectations on frontier-AI cyber risk under DORA. |
| Israel (BoI / ISA / CMISA) | Israel (Banking Supervision Department: banking corporations and payment service providers with prudential importance; ISA: investment advisers, marketers and portfolio managers; Capital Market Authority: insurers and savings) | Supervisory guidance | Proper Conduct of Banking Business Directive 369, Management of Model Risk (Bank of Israel, 21 August 2024, in effect 21 August 2025), read with the inter-agency taskforce's final report on AI in the financial sector (December 2025) | 4 Nov 2026: the Israel Securities Authority published a draft directive on services provided through technological means and AI for investment advice and portfolio management licensees, with comments due by 6 December 2026. |
| UAE (CBUAE / DFSA / FSRA) | United Arab Emirates (CBUAE: banks, finance companies, insurers and other licensed financial institutions onshore; DFSA: Authorised Firms in the Dubai International Financial Centre; FSRA: firms licensed in Abu Dhabi Global Market; Capital Market Authority: capital-market firms) | Supervisory guidance | Guidelines for Financial Institutions adopting Enabling Technologies (CBUAE, SCA, DFSA and FSRA, 15 November 2021), with the CBUAE Guidance Note on AI and machine learning (11 February 2026) onshore and the DFSA's letter on AI risk management (4 June 2026) in the DIFC | 4 Nov 2026: at its supervisory roundtable with bank and insurer CEOs the CBUAE set out its 2027 regulatory priorities and held a session on thematic supervisory reviews in banking, particularly digital assets and AI; 2 Nov 2026: the DFSA published its third annual AI survey (71% of responding firms use AI, 53% exploring to deploying AI agents); 30 Oct 2026: the DFSA and C:>DIR published an insights paper on frontier and agentic AI and cyber resilience. |
| Qatar (QCB) | State of Qatar (QCB: all QCB-regulated financial institutions; QFCRA: firms authorised in the Qatar Financial Centre) | Binding law | Artificial Intelligence Guideline (Regulating the use of Artificial Intelligence by QCB Licensed Entities), in force from 4 September 2024 | February 2025: the QCB announced its Data Handling and Protection Regulation, which requires QCB approval before data is processed in fully autonomous decision-making systems and lists the AI Guideline among the secondary regulations institutions must comply with. As of 19 November 2026 this review found no later AI-specific instrument on the QCB website; the AI Guideline of 4 September 2024 remains the operative text. |
| Saudi Arabia (SAMA / SDAIA) | Kingdom of Saudi Arabia (SAMA: banks and other institutions it supervises; SDAIA: cross-sector AI and data governance for government and private entities) | Voluntary framework | SAMA Counter-Fraud Framework (Circular No. 44021528, 11 October 2022) for the binding AI control on banks, read with SDAIA's AI Ethics Principles (Version 1.0, September 2023; 2025 edition SDAIA-P114E) for cross-sector AI governance | 15 Sep 2026: SAMA's Governor told Money20/20 Middle East that advances in AI have brought finance to 'the frontiers of a new era in finance, one which is powered by AI agents, real-time data, and fluid cross-border activity', which must be addressed 'with care and deliberation'; 14 Jul 2026: SDAIA launched the advisory National AI Risk Management Framework for government and private entities (English executive summary SDAIA-P145EN, dated April 2026); 8 May 2026: the Saudi Press Agency reported that SDAIA had issued its Deepfakes Guidelines, which list financial fraud and impersonation among the main risks. |
| South Africa (FSCA / PA) | South Africa (FSCA: market conduct of financial institutions; PA: prudential supervision of banks, insurers and market infrastructures) | Supervisory guidance | Artificial Intelligence in the South African Financial Sector (FSCA and Prudential Authority, 24 November 2025) | 6 Aug 2026: FSCA Commissioner Unathi Kamlana said the FSCA, the SARB and the PA are developing a joint discussion paper on AI in the financial sector, including agentic AI, and intend to add high-level AI governance principles to the Joint Standard on culture and governance requirements. |
| China (NFRA / PBoC / CAC) | People's Republic of China (mainland): NFRA supervises banks and insurers, the PBoC sets fintech policy and financial technology standards, the CAC regulates algorithms and generative AI services across all sectors | Binding law | NFRA Measures for Data Security Management of Banking and Insurance Institutions (金规〔2024〕24号, 27 December 2024), read with the CAC Interim Measures for the Management of Generative AI Services (in force 15 August 2023) | 1 Sep 2025: the CAC's Measures for Labelling AI-Generated and Synthetic Content took effect (issued 7 March 2025, published 14 March 2025); 27 Dec 2024: the NFRA issued its Data Security Measures with AI model governance provisions for banks and insurers. |
| South Korea (FSC / FSS) | Republic of Korea (FSC and FSS: banks, financial holding companies, financial investment firms, insurers, card companies, savings banks, electronic financial service providers and other supervised institutions; Ministry of Science and ICT: cross-sector duties under the AI Basic Act, in force since 22 January 2026) | Binding law | AI Guidelines in the Financial Sector (FSC, announced 18 June 2026, in effect 22 June 2026), read with the Framework Act on the Development of Artificial Intelligence and the Creation of a Foundation for Trust (Act No. 20676, in force 22 January 2026) | 18 Nov 2026: the FSC proposed amendments to the supervisory regulation on electronic financial services (up-to-date IT asset registers, third-party vulnerability evaluations every six months for large firms, emergency inspections on demand), with comments to 30 November 2026 and effect planned for the first quarter of 2027, following its 5 November plan against AI-driven hacking; 3 Sep 2026: second phase of the temporary easing of network separation for frontier-AI cyber testing (75 eligible companies, 15 to be selected). |
| Malaysia (BNM / SC) | Malaysia (BNM: licensed banks, Islamic banks and investment banks, insurers and takaful operators, prescribed development financial institutions, e-money issuers and other payment firms; SC: capital market entities) | Supervisory guidance | Feedback Statement: Artificial Intelligence (AI) in the Malaysian Financial Sector (BNM, 19 November 2026), read with the August 2025 Discussion Paper and the binding RMiT, Fair Treatment of Financial Consumers and e-KYC policy documents | 19 Nov 2026: BNM published its Feedback Statement on AI in the Malaysian Financial Sector (113 respondents), confirming it is not introducing AI-specific regulation at this stage, mapping AI onto existing policy documents, setting out emerging practices for agentic AI and announcing an AI Responsible Innovation Forum and AI testing pathway to be explored with industry from 2027; 25 Sep 2026: RMiT policy document updated; 8 Jul 2026: Governor's AICB Nexus speech: 'Responsibility cannot be delegated to an algorithm'. |
| Indonesia (OJK) | Indonesia (commercial banks and other financial services firms supervised by OJK, including fintech providers; Bank Indonesia separately regulates and supervises payment system operators) | Supervisory guidance | Artificial Intelligence Governance for Indonesian Banking (Tata Kelola Kecerdasan Artifisial Perbankan Indonesia), launched by OJK on 29 April 2025 | 2 Oct 2026: OJK launched the Roadmap for the Development and Strengthening of Indonesian Banking (RP3I) 2026-2030, which says OJK will issue an AI Maturity Index assessment of banks and treats frontier AI models as magnifying the speed and scale of cyber risk; 1 Mar 2026: PADK No. 1 of 2026 on IT implementation by commercial banks took effect, revoking SEOJK No. 21/SEOJK.03/2017. |
| Bank of Thailand | Thailand (financial institutions and specialized financial institutions under the Financial Institutions Businesses Act B.E. 2551 (2008); payment system operators and payment service providers under the Payment Systems Act B.E. 2560 (2017)) | Supervisory guidance | Policy guideline on risk management of the use of artificial intelligence systems (BOT circular ธปท.ว. 5994/2568, 12 September 2025; published in Thai) | 14 Sep 2026: the BOT issued minimum practices for digital fraud management (ธปท.ว. 5713/2569) requiring domestic systemically important banks and significant e-money providers to develop and start using technology-based abnormal-transaction detection within 2026, with AI and machine learning named as examples; 12 Sep 2025: the BOT issued its AI risk management policy guideline (ธปท.ว. 5994/2568) after consulting from 12 to 30 June 2025. |
| Philippines (BSP) | Philippines (BSP-supervised financial institutions, BSFIs: banks, non-bank financial institutions with quasi-banking functions, e-money issuers and other payment providers) | Supervisory guidance | Memorandum No. M-2026-031: Governance Principles for Artificial Intelligence (AI) in Financial Services (24 June 2026) | 24 Jun 2026: the BSP issued Memorandum No. M-2026-031 with the voluntary Governance Principles for AI in Financial Services (STARS); the model risk management circular proposed in June 2025 had not been issued in final form as of that memorandum. |
| Brazil (BCB / CMN / ANPD) | Brazil (BCB and CMN: banks and other institutions licensed by the Central Bank; ANPD: anyone processing personal data under the LGPD) | Binding law | CMN Resolution 4,893 (26 February 2021, amended by CMN Resolution 5,274 of 18 December 2025) on cyber security and the contracting of data processing, storage and cloud computing services | 18 Dec 2025: CMN Resolution 5,274 amended Resolution 4,893 with minimum cyber security controls and Pix and STR isolation; Nov 2025: the BCB published its survey on AI use in the National Financial System in the Financial Stability Report; PL 2338/2023 awaits the rapporteur's opinion in the Chamber's special committee (status checked 2026-11-20). |
The EU deferred its high-risk AI deadline. The high-risk regime — including credit scoring — was due August 2, 2026; the Digital Omnibus on AI (Regulation (EU) 2026/1744, in force July 27, 2026) moved it to December 2, 2027. The obligations (risk management, data governance, human oversight, logging; fines up to €15M or 3% of global turnover) are unchanged. Details →
The CFPB narrowed fair-lending exposure for AI models. Its April 2026 Regulation B rule (effective July 21, 2026) says ECOA does not authorize disparate-impact liability — while the duty to give specific adverse-action reasons is untouched, and its 2022/2023 AI circulars were withdrawn in May 2025. Details →
The US retired SR 11-7. On April 17, 2026 the OCC, Federal Reserve, and FDIC replaced the 2011 model risk management framework — and explicitly excluded generative and agentic AI from its scope, leaving those to banks' broader governance programs. Details →
The OCC gave its examiners an AI framework. On November 4, 2026 it issued Bulletin 2026-54, “Super Intelligence: Supervisory Framework” — sixteen components, six pillars and a four-tier, consequence-based classification for the generative and agentic AI that the April guidance excluded, under the “Super Intelligence” name of Executive Order 14434. It sets no new expectations for banks; it tells examiners what to look at. Details → In plain English →
The FSB finalised firm-level expectations. On October 15, 2026 it published the final report on 12 sound practices for responsible AI adoption — including for agentic AI — closing out its June 2026 consultation. Details →
| Date | Authority | Development |
|---|---|---|
| Oct 7, 2028 | MAS | MAS AI Risk Management Guidelines: lifecycle-control expectations (Sections 5-6) apply. Second phase of the transition in the final Guidelines on Artificial Intelligence Risk Management: the lifecycle-control and capability expectations in Sections 5 and 6 are to be met by 7 October 2028, a year after the oversight and risk-management-system expectations in Sections 3 and 4 (7 October 2027, the document's effective date). |
| Aug 2, 2028 | EU AI Act | Deferred deadline for high-risk AI embedded in regulated products. Article 6(1) / Annex I high-risk systems tied to EU product-safety legislation now apply from this date (moved from August 2, 2027 by the Digital Omnibus on AI). |
| Jan 13, 2028 | Taiwan (FSC / MODA) | End of the two-year period for Taiwan's government to adapt existing laws to the AI Basic Act. Article 18 gives the government two years from the Act's effective date to enact, amend or repeal laws and administrative measures that do not comply with it. The Ministry of Digital Affairs' October 2026 report to the Legislative Yuan puts the deadline at 13 January 2028 and says eight agencies, including the Financial Supervisory Commission, are expected to complete the risk framework's process by the end of 2026. |
| Dec 2, 2027 | EU AI Act | High-risk AI obligations for stand-alone Annex III systems become applicable. Deferred from August 2, 2026 by the Digital Omnibus on AI. Annex III high-risk systems — including credit scoring of natural persons — must comply with risk management, data governance, documentation, logging, human oversight, accuracy, and post-market monitoring requirements. Fines up to €15M / 3% of turnover. |
| Dec 2, 2027 | BaFin | BaFin begins monitoring high-risk AI system requirements. BaFin states that from 2 December 2027 it will also monitor compliance with the requirements for high-risk AI systems, which under Annex III(5)(b) and (c) of the AI Act include systems evaluating the creditworthiness or credit score of natural persons (except fraud detection) and risk assessment and pricing for life and health insurance. |
| Aug 2, 2027 | EU AI Act | Deadline for general-purpose AI models placed on the market before August 2025. Providers of general-purpose AI models placed on the market before 2 August 2025 must comply with the AI Act's GPAI obligations by this date. |
| Jul 10, 2027 | EBA | EU Anti-Money Laundering Regulation applies. Regulation (EU) 2024/1624 (AMLR) applies from 10 July 2027, setting the customer due diligence rules that remote and AI-assisted onboarding under EBA/GL/2022/15 must meet. |
| Jul 1, 2027 | Utah Division of Consumer Protection | Scheduled repeal of the Artificial Intelligence Policy Act (Title 13, Chapter 72). Section 63I-2-213 provides that Title 13, Chapter 72, Artificial Intelligence Policy Act, is repealed July 1, 2027, the date SB 332 (2025) substituted for May 1, 2025, unless the 2027 General Session acts on the Business and Labor Interim Committee's September 16, 2026 recommendation to extend it to July 1, 2037. The listed repeal does not name Chapter 77, the generative-AI disclosure chapter. |
| Jan 1, 2027 | State of California | CPPA ADMT obligations for significant decisions apply. The CPPA package approved on September 22, 2025 took effect January 1, 2026 (see the regulations document); its ADMT requirements for significant decisions (pre-use notice, opt-out or human appeal, access to an explanation) apply from January 1, 2027. Risk-assessment attestations and the first cybersecurity-audit certifications are due April 1, 2028. |
| Jan 2027 | Central Bank of Ireland | Applications open for the Advancing AI for Consumer Good sandbox. The Central Bank's third Innovation Sandbox Programme runs on a cohort basis over six months and gives participants structured regulatory advice and support on four problem statements: consumers in vulnerable circumstances, fairness and explainability in automated decisions, resilience to fraud and scams, and accountability as AI moves from advising to acting. It is not an AI Act regulatory sandbox and offers no derogation or waiver from regulation. |
| Dec 2, 2026 | EU AI Act | Article 50(2) marking transition ends. The Digital Omnibus transition period for Article 50(2) machine-readable marking of AI-generated content ends; the rest of Article 50 has applied since 2 August 2026. |
| Dec 1, 2026 | Colorado AI Act | Regulation 10-1-1: annual life insurer reports and non-use attestations due. Life insurers file their annual SB 21-169 governance report, and insurers that do not use external consumer data or algorithms file attestations, under the Division of Insurance's amended regulation. |
Who regulates AI in banking?
No single body does. Banks face a layered system: binding law where it exists (the EU AI Act, US consumer statutes like ECOA), prudential supervisors applying existing frameworks to AI (Federal Reserve, OCC, ECB, UK PRA/FCA), and global standard-setters shaping the agenda (FSB, Basel Committee). Voluntary frameworks like the NIST AI RMF fill the gaps supervisors leave open.
Is there a dedicated AI law for banks?
At the federal level, only in the EU: the EU AI Act is the sole binding, cross-sector AI law that reaches banks, and its high-risk regime covering credit scoring applies from December 2, 2027 after the 2026 Digital Omnibus deferral. In the US there is no federal AI statute for banks — but two states bind them: Colorado's Automated Decision-Making Technology Act (effective January 1, 2027) reaches lenders with no bank exemption, and California's CPPA regulations on automated decisionmaking technology (obligations from January 1, 2027) list lending as a significant decision, though GLBA-covered data is exempt. The UK has deliberately chosen to regulate bank AI through existing law and supervision.
What changed for bank AI regulation in 2026?
Five big moves: the US agencies replaced the 15-year-old SR 11-7 model risk framework with revised guidance that excludes generative and agentic AI (Apr 17); the CFPB's Regulation B rule eliminated disparate-impact liability under ECOA (effective Jul 21); the EU's Digital Omnibus deferred the AI Act's high-risk regime, including credit scoring, to December 2, 2027; the FSB finalised its 12 sound practices for responsible AI adoption on October 15, 2026, as a G20 deliverable; and on November 4 the OCC issued Bulletin 2026-54, a Super Intelligence supervisory framework telling its examiners how to assess generative and agentic AI through sixteen components and four consequence-based tiers, without setting new expectations for banks.
Regulators move daily. So do we.
when one of these regulators moves, the next morning's brief says so · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning