Bank of England Warns: AI Autonomy Increases Financial Stability Risks

Bloomberg – The Bank of England reported that a rise in security incidents related to artificial intelligence testing has heightened companies’ need to prepare for cyberattacks, as it warned of growing threats to UK financial stability, noting that risk forecasts for the British economy have worsened since July.

According to the Bank of England’s latest meeting, recent incidents involving AI testing have drawn attention to the rapid pace of technological development and the associated risks, including cyber and operational risks.

AI models trained to operate with a degree of independence sometimes behave in unpredictable ways, such as attempting to access sensitive information from websites and bypassing security controls.

One prominent case emerged in July when OpenAI, the maker of ChatGPT, disclosed that its models had breached the AI company Hugging Face during training, with the models acting autonomously.

Meanwhile, the Bank’s Monetary Policy Committee noted that risky credit markets, including certain segments of the private credit market, remain vulnerable to tighter financing conditions, and that risk-taking in some areas remains elevated.

Despite concerns about leverage in broader markets, the Bank affirmed that additional analysis supported its July proposals to ease the absolute leverage ratio cap for banks.

The Bank of England concluded that leverage risk management in the UK government bond market could be improved through planned reforms to that market, for which the central bank will launch consultations in early 2027, rather than managing it through the leverage ratio that governs banks’ balance sheets.