Bank of England: Right to Intervene in AI as Cyber and Debt Risks Grow
Bank of England Governor Andrew Bailey says authorities must keep the "right to intervene" in the artificial intelligence industry in case rogue models threaten the financial system. He described the risks from rapid advances in frontier AI models, the most powerful cutting-edge systems, as "real and increasingly significant." The comments came in the first opinion piece for the Bank of England's Insight series. The Guardian
Bailey said AI has increased the "scale and sophistication of cyber threats to the financial system." That could affect card payments, bank transactions and stock and bond trading. The Guardian The warning follows his August 2026 letter as Chair of the Financial Stability Board, the international body that monitors financial risks, to G20 Finance Ministers and Central Bank Governors. In it, he said the potential impact of frontier AI on cyber risk is the most immediate concern to the financial system. FSB
Bailey said regulation is not the right place to start on AI. He warned against moving too quickly to debates about regulatory architecture, or who regulates what, before it is clear where failure could happen. He proposed rigorous testing of new AI models as a "sensible starting point" to understand complex systems and find credible points for intervention. The Guardian Testing comes first. Regulation does not.
The Bank of England's Financial Policy Committee, which watches risks across the whole system, warned that growing AI-related debt is adding to financial stability risks. It said large AI players took on $450bn (£339bn) in debt between January and September 2026. That exceeds the $333bn in UK gilts, or government bonds, due for issuance in all of 2026. The Guardian
Bailey has previously urged a measured stance. In October 2025 he said the UK should take a 'pragmatic approach' to AI and called for a supportive domestic environment to help investment in AI and other projects that need time to pay off. Reuters In September 2025 he said AI can help regulators find the 'smoking gun'. Reuters
In September 2026, Bailey delivered a keynote speech at the LSE TRIUM anniversary conference, where he argued that independent central banks rest on more than technical expertise alone. Bank of England He gave a speech on growth and regulation at the Mansion House Financial and Professional Services Dinner in London. BIS That same month he also said weak productivity and shocks such as COVID-19 were behind a climb in public debt. Reuters
The broader context here is a central bank trying to keep the option to act without locking in how it would act. Bailey presents testing as a way to learn where pressure can usefully be applied inside systems that are hard to see into, rather than writing rules for failures that are not yet mapped. For supervisors used to checking models, holding capital buffers and setting standards for coping with disruption, that applies familiar thinking to unfamiliar technology.
Looking at what this means for policymakers, three tensions will be difficult to manage. First, cyber risk is immediate and crosses borders, which pushes coordination toward the Financial Stability Board and G20 group Bailey already chairs. Second, AI-related borrowing is now so large it can be compared with government borrowing, which raises questions about market liquidity and exposure to other firms if confidence shifts. Third, the call for a supportive investment climate sits alongside the call for a right to step in, and markets will watch how long testing stays a tool for learning before it becomes a tool for setting rules.
