UK regulators publish effective practices for AI cyber risk and resilience
For the CISO of a US regional bank
The Bank of England, the Prudential Regulation Authority (PRA) and the Financial Conduct Authority (FCA) jointly published effective practices on cyber risk and resilience management relating to AI on Nov. 19.1
The regulators said AI is rapidly creating new cyber risks and amplifying existing ones, while also letting firms strengthen their cyber defences.1 The FCA said the practices build on the regulators' cyber guidance, which is out for consultation, and help firms prepare for specific AI and quantum risks.3
The regulators said the practices are separate from their supervisory statement and final guidance on cyber risk and resilience management, add detail on AI, and are not part of formal expectations.1
According to the publication, advanced AI can analyse source code, find vulnerabilities and discover attack paths, letting threat actors exploit flaws within hours or days rather than weeks or months.1
The publication says firms need to evolve from traditional vulnerability management, where discovery has historically been the main bottleneck, towards continuous exposure management, and says effective practice may also require firms to increase execution speed: how fast they identify affected assets, test remediation, secure a change window, coordinate with third parties, deploy remediation and verify its success.1
The regulators said advanced AI systems are non-deterministic and may give different results across repeated assessments, so outputs need validation, and appropriate human oversight remains essential for key decisions; firms may scale the practices to their reliance on AI and to the autonomy and access of the AI systems they use.1
Supervisors are signaling that patching and remediation speed, not detection, is where AI-era cyber risk will be tested. Banks should check how fast their change windows and third-party coordination run, and keep human sign-off on automated security decisions, even though the practices are not formal expectations.
The FCA said the AI and quantum material is not open for formal consultation, but the related cyber guidance consultation closes on Feb. 26, 2027, with final guidance due next year.3
Our take. UK regulators say AI lets attackers exploit flaws in hours or days, not weeks or months, yet they still want humans on key defence decisions. It binds nobody. It is still a free answer key for your next exam.
Action. Information security: Measure days from vulnerability disclosure to verified patch on critical systems, and name who approves emergency changes.
What our records say
The UK has not written an AI rulebook for finance: the Bank of England, PRA and FCA apply existing technology-agnostic frameworks (senior manager accountability, model risk management (SS1/23), operational resilience and the Consumer Duty) to AI, a position most recently reaffirmed in April 2026. PRA SS1/23: It sets five principles (model identification and risk classification, governance, development and use, independent validation, and risk mitigants), covering all models used to inform business decisions, including vendor models, regardless of technology.12
From our records
- UK (BoE / PRA / FCA)regulator · binding Supervisory guidance · latest 2026 BoE/FCA AI survey (Oct 2026), FCA: AI in financial services — good practice and areas for improvement (Oct 2026)
Sources
- Primary sourceEffective practices: Cyber risk and resilience management relating to artificial intelligence (AI) | Bank of England – the UK's central bank Bankofengland
- Primary sourceThe AI shift in cyber risk: why leaders must act now | National Cyber Security Centre Ncsc
- Primary sourceUK financial watchdogs clarify cyber expectations as attacks rise Fca