AI in action · Regulation
BoE's Bailey Names Two AI Risks: Cyberattacks and a Bubble
Bank of England Governor Andrew Bailey said on November 9, 2026, that frontier AI has raised cyber risk and that an AI share bubble could hurt stability.
Andrew Bailey, Governor of the Bank of England, chose a university lecture rather than a policy meeting to set out how AI could threaten financial stability. Speaking on Monday, November 9, in a series run by the Swiss Institute of International Studies at the University of Zurich, he named two channels, according to the London Evening Standard: cyberattacks made more dangerous by frontier models, and a share-price bubble in AI companies.
On the first, he was direct. "In our world, frontier AI has increased the risk from cyber-attacks – something that was already prominent," Bailey said, adding that the concerns "may call into question public trust in AI. This is critical." His remedy was more scrutiny of the models themselves: "more extensive model testing" before and after release, and "better incident reporting, monitoring and follow up" when AI tools go wrong.
On the second, he described two ways the market could be disappointed. Investors may simply be overvaluing AI companies, or the returns may be real but slower to arrive than prices assume; in his words, "the timing may not live up to market expectations." Either outcome could hit UK financial stability through the banks, insurers and funds that have lent to or invested in the build-out.
Bailey did not sound like an AI pessimist. He called himself an optimist and said AI, digital technology and "most probably robotics too" could raise growth through higher productivity: "That is a good thing." The Bank's own Monetary Policy Report the previous week had been more cautious, estimating that AI had added about 0.5% to UK productivity over three years.
The lecture came days after the UK's National Cyber Security Centre warned in its annual review that AI-boosted attacks on critical infrastructure could arrive within two years. For financial firms, the incident-reporting call is the practical part: it points to what supervisors may soon expect from banks and insurers using AI, and keeping a record of model changes and AI incidents now costs little compared with building one under a deadline later.
Sources: London Evening Standard — Andrew Bailey on frontier AI, cyber risk and an AI share bubble (Nov 2026); The Register — Bank of England pours cold water on AI productivity boom (Nov 6, 2026)