October saw the second-highest monthly number of cyber incidents reported to the Financial Conduct Authority in five years, the Bank of England said on Thursday (Bank of England). Reuters put the total at 38 (Reuters).
That is 52% above the previous 18-month peak, recorded in May 2025. The regulators now say AI lets attackers exploit flaws “within hours or days rather than weeks or months”.
Business Pill · Least Privilege
An evergreen Business Engineer explainer on least privilege and access control. It teaches the general idea only and does not discuss UK regulators or these incidents.
The key insight: UK firms reported 38 cyber incidents in October, 52% above the May 2025 peak, and regulators say AI lets attackers exploit flaws within hours or days. Their AI paper adds that results in cyber defence may depend as much on the harness as on the model.
What the Regulators Published
The Bank of England, the FCA and the Prudential Regulation Authority opened a consultation on new cyber resilience guidance for financial firms. It closes on 26 February 2027.
The guidance covers how firms govern, identify, protect against, detect, respond to and recover from cyber incidents. The regulators say it brings together existing requirements rather than imposing rigid rules.
Separately, they published “effective practices” on cyber risks linked to AI and quantum computing (AI effective practices). That material is not open for formal consultation.
The Financial Times reported the move under the headline “City firms told to tighten cyber defences as AI exposes vulnerabilities” (FT).

What Was Posted on X
Rohan Paul summarised the Financial Times report and attached a screenshot of the FT headline. His post is the source for the ‘bare-metal’ detail, which we could not confirm in the regulators’ own papers; they speak of immutable backups and manual fallbacks.
How Big the Jump Is
October’s 38 incidents were surpassed only by the 49 reported in June 2023, during the MOVEit data breach, according to Reuters.
If 38 is 52% above the May 2025 peak, that peak was about 25 incidents, by our arithmetic. The regulators did not publish the May figure in the release.
The Bank also cites its latest Systemic Risk Survey, which named cyber as one of the risks with the greatest impact on the UK financial system.
What AI Changes, in the Regulators’ Words
The AI paper says advanced systems “can analyse source code, identify vulnerabilities and discover attack paths at great speed”.
It says firms should evolve from traditional vulnerability management to continuous “exposure management”. It adds that “firms need to treat remediation speed and vulnerability backlogs as operational risks”.
It also names a defensive point we have tracked: “The number of vulnerabilities a firm discovers and remediates may depend as much on the harness as on the underlying AI.” We covered Ramp’s benchmark on harnesses on Thursday.
Recovery plans, the paper says, may need to consider “AI agents gaining unauthorised access to recovery environments”. Options include “manual fallbacks and immutable backups”.
Update, 19 November 2026, 13:52 UTC: Germany’s cyber agency, the BSI, has raised its threat rating to “tense to critical” and names AI as a reason. The Federal Office for Information Security gave the rating in its annual report, its press release says. Last year it called the situation “tense”, Reuters reported.
Besides the geopolitical situation, the reason for the reassessment is the rapid development of AI, the BSI says, in our translation.
The time from a flaw becoming known to its first known exploit averaged 53 days in the previous reporting period and 25 days in the current one. In the second quarter of 2026 it was 8 days, a fall of 85%, the BSI says.
About 160 new vulnerabilities became known every day in the period, 35% more than a year earlier, and the agency’s new BSI Cyber Index averaged 170 points, up 50%, the BSI says.
BSI president Claudia Plattner said the agency is dealing with tools “that enable cyberattacks in previously unknown volumes and speeds”, in our translation. The report said email-based attacks were declining as attackers used AI models to attack systems directly, Reuters reported.
That echoes the UK regulators above: exploits “within hours or days rather than weeks or months”. By our arithmetic, Germany’s Q2 figure is about 15% of the previous period’s 53 days, matching the BSI’s 85% fall.
The Survey Numbers
In a Bank of England and FCA survey cited in the release, 88% of respondents expected frontier AI cyber capabilities to have a moderate or high impact on their risk profile over the next two years.
In the same survey, 64% said vulnerability management and patching had seen the greatest increase in management attention due to awareness of frontier AI.

The Losses Behind It
The FBI said on 31 October that North Korea was responsible for the theft of more than $6 billion in virtual assets from Binance on or about 29 October (FBI).
Reuters also cited French lender Credit Mutuel Arkéa, which said hackers had stolen around €100 million in a cyber heist earlier this month.
PRA chief executive Katharine Braddick said: “Cyber attacks are becoming more frequent, complex and dangerous.” FCA chief executive Nikhil Rathi said: “Our response shouldn’t always be more rules.”
The Structural Read
As we read the AI paper, the regulators are shifting the test from whether firms have controls to how fast they fix and recover: remediation speed and vulnerability backlogs become operational risks.
As we read the recovery section, AI agents are now a named threat to the backup layer itself, which is why the paper lists immutable backups and manual fallbacks.
Katharine Braddick, CEO of the PRA and Deputy Governor at the Bank of England
“Regulators need confidence both that firms have strong cyber controls and that they can detect, contain and recover quickly from incidents.”
Three Implications
FOR BANKS AND INSURERS The consultation on cyber resilience guidance closes on 26 February 2027, per the Bank’s release.
FOR SECURITY TEAMS The AI paper asks firms to treat remediation speed and vulnerability backlogs as operational risks.
FOR AI VENDORS The paper names cyber-attacks by highly autonomous AI agents among the attacks firms should plan to recover from.
What Is Not Established
Rohan Paul’s summary of the FT report on X says regulators told firms to build offline “bare-metal” systems for rebuilding after hacks (post). We could not read the FT article; the regulators’ papers we read speak of immutable backups and manual fallbacks.
The Bank’s release does not say how many October incidents involved AI, and it does not give a breakdown by firm type.
We covered Arizona’s probe of OpenAI over rogue AI agents this week; the UK papers do not name any company.
The Bottom Line
UK firms reported 38 cyber incidents in October, 52% above the prior 18-month peak. Regulators opened a consultation to 26 February 2027 and told firms AI lets attackers exploit flaws within hours or days.
97,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The 38 count and the June 2023 comparison are from Reuters; the 52% rise, quotes, the consultation and the survey figures are from the Bank of England release and the regulators’ AI effective-practices paper; the Credit Mutuel Arkéa figure is also from Reuters; the Binance attribution is from the FBI. The FT is cited by headline only; we could not read its article, and the ‘bare-metal’ detail is from a summary on X. The implied May 2025 figure is our arithmetic. Not investment advice; for information only.
Sources: Bank of England news release, 19 Nov 2026 · Bank of England / PRA / FCA, Effective practices: cyber risk and AI, Nov 2026 · Reuters via KFGO, 19 Nov 2026 · FBI alert, 31 Oct 2026 · Financial Times, headline as shown on X, 19 Nov 2026 · Rohan Paul on X, 19 Nov 2026









