The Bank of England plans to revise its approach to regulating artificial intelligence in the financial sector. This was stated by Deputy Governor Sarah Breeden at the European Central Bank forum in Sintra. According to her, current technology‑neutral rules were designed without considering systems capable of operating without direct human oversight, and the main risks of mass adoption of AI agents are cyberattacks, payment system failures, and synchronous errors that could destabilise markets. Breeden stressed that the accelerating development of AI creates a dual challenge: not only ensuring responsible adoption and risk management, but also adapting the working methods of central banks themselves to the new reality.

Among possible regulatory measures, Breeden mentioned tighter requirements for critical system recovery, fully isolated backup capacity, and mechanisms for rapid reconfiguration of compromised infrastructure. For financial markets, the regulator is exploring safeguards akin to trading circuit breakers in case flawed AI models trigger large‑scale disruptions. Particular attention is being paid to the risk of herding behaviour, where autonomous agents react to the same market triggers and amplify volatility. These initiatives come against a backdrop of international warnings: on 15 May, the Bank of England, the Financial Conduct Authority and HM Treasury issued a joint statement on the cyber risks of advanced AI models; on 22 June, a similar assessment was made by the Five Eyes alliance; and on 10 June, the Financial Stability Board published a consultation paper outlining 12 practices for responsible AI adoption by financial institutions.

According to the Cambridge Centre for Alternative Finance, 81% of surveyed industry participants already use AI at some level, while 52% are at the pilot or more mature deployment stage of agentic systems. So far, the main application remains internal processes: automation, data visualisation, and software development, while front‑office adoption remains limited. In a response to the parliamentary committee on 1 April, the Bank of England stated that generative and agentic AI is not yet being used in the financial system on a scale that could create systemic risk, but stressed that the technology‑neutral regulatory approach will be reviewed. In May 2026, the International Monetary Fund called for AI‑related cyber risks to be treated as a financial stability issue rather than merely an IT security concern, and the discussion continues at other institutions, including the Bundesbank and the Bank for International Settlements.