The Bank of England's new cyber rules assume banks will get hit — by AI, at machine speed
Reported cyber incidents jumped 52% above the previous peak in October. The regulators' answer is guidance on AI and quantum risk, not new rules.
- Reported cyber incidents in October 2026 were 52% above the previous 18-month peak recorded in May 2025, the second-highest monthly number reported to the FCA in five years. m:…9954
- 88% of survey respondents expect frontier AI cyber capabilities to have a moderate or high impact on their risk profile over the next two years. m:…9954
- A 2025 survey of experts suggests a cryptographically relevant quantum computer is likely to be developed within 15 years, with a meaningful probability within 10 years. m:…0932
- The joint consultation paper CP17/26 and GC26/9 does not propose new rules; it clarifies how firms should meet current requirements, and the consultation closes on 26 February 2027. m:…0934
What the regulators published
The Bank of England, the Financial Conduct Authority and the Prudential Regulation Authority have proposed joint guidance to help financial services firms prepare for and recover from cyber incidents. The consultation paper, CP17/26 and GC26/9, does not propose new rules. Instead it clarifies how firms should meet current requirements in governance, operational resilience, operational risk and third-party risk management. The proposed guidance covers governance, identification, protection, detection, response and recovery, and it aligns with the Cyber Security Centre's Cyber Assessment Framework. Alongside the consultation, the regulators published two non-binding effective practices documents, one for managing cyber risks relating to artificial intelligence and one for quantum computing. The consultation closes on 26 February 2027, and the regulators say they welcome feedback from firms and financial market infrastructures on how the effective practices documents could be improved. m:…9954 m:…0934
Why now: the incident count
The timing follows a sharp rise in reported incidents. October 2026 saw the second-highest monthly number of cyber incidents reported to the FCA in five years, and reported incidents were 52% above the previous 18-month peak recorded in May 2025. The regulators' own survey found that 88% of respondents expect frontier AI cyber capabilities to have a moderate or high impact on their risk profile over the next two years, and 64% said vulnerability management and patching had seen the greatest increase in management attention because of frontier AI awareness. The AI effective practices document explains why: AI systems are significantly reducing the cost, skills and time required for threat actors to exploit vulnerabilities, and advanced AI systems can analyse source code and identify vulnerabilities at speeds that allow exploitation within hours or days. The UK AI Security Institute reports that AI performance in some areas is doubling every eight months. m:…9954 m:…0933
The document's prescription is a change in posture. Firms need to evolve from traditional vulnerability management to continuous exposure management, and should assume they will suffer disruption and design systems to contain and recover from machine-speed attacks. Cyber risk is a core business risk and a leadership responsibility, the document says, and firms should implement layered controls and human oversight for highly autonomous AI systems. The stakes go beyond individual firms: AI-related cyber risks could threaten financial stability if they affect shared resources used by many firms. The document also points to collaboration through groups like CMORG and FAISF as essential for responding to machine-speed attacks, and says effective harness engineering is key to using AI for cyber defence. m:…0933
The quantum deadline
The quantum document looks further ahead and asks boards to start now. A cryptographically relevant quantum computer would be able to solve mathematical problems that classical computers cannot, undermining current public-key cryptography — the cryptography that underpins financial transactions. Threat actors may already be using a harvest now, decrypt later strategy, intercepting encrypted data for future decryption. A 2025 survey of experts suggests such a computer is likely to be developed within 15 years, with a meaningful probability within 10 years. The document puts boards and senior management in charge of setting the strategic direction of quantum migration programmes, and tells firms to maintain a cryptographic inventory, build crypto-agility into systems so algorithms can be replaced, and remember they remain accountable for quantum-related cyber risk even when systems are provided by third parties. m:…0932 m:…9954
The market is already there
The vendor market is moving in the same direction. Microsoft Perception, an agentic security system that orchestrates red, blue and green team agents through playbooks in a closed loop, is generally available as of November 2026, and Microsoft classifies AI agents as a new insider-risk category. Darktrace, writing about the launch, says the window between vulnerability discovery and exploitation is collapsing from months to minutes, and that behavioural defence is required to identify actions that are allowed by policy but abnormal for an agent. Microsoft's security foundation produces more than 165 trillion signals per day. The implication for banks is that the attacker's tempo is now set by machines, and the regulators' guidance is built around that fact. m:…9509
What is still unresolved
- Whether guidance without new rules changes how firms actually spend on cyber resilience, or whether it is treated as a restatement of existing expectations.
- When a cryptographically relevant quantum computer will actually arrive — the expert survey gives a range of 10 to 15 years, not a date.
- Whether shared-resource concentration means a single AI-driven incident could become a systemic event, as the regulators warn it could.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.