Andrew Bailey, chair of the Financial Stability Board (FSB), the international body that monitors vulnerabilities across the global financial system, warned on Monday that frontier artificial intelligence poses the most immediate threat to financial stability worldwide. The warning came in a letter published ahead of the G20 meeting in Asheville, North Carolina.
Bailey, who also serves as governor of the Bank of England, told G20 finance ministers and central bank governors that AI systems capable of operating with growing autonomy represent a qualitative shift in the threat environment for financial institutions.
"Frontier AI may have the ability materially to alter the speed, scale and economics of cyber risk, which could undermine market confidence system-wide, especially due to highly concentrated third-party service providers," Bailey wrote.
The concern is specific. A small number of hardware, cloud and model providers supply the technology infrastructure on which much of the global financial system now depends. If frontier AI can identify and exploit vulnerabilities at machine speed and scale, failures need not originate inside a bank or exchange to cascade through one.
Bailey said frontier AI models are demonstrating growing autonomy and more capable problem-solving, alongside an expanding range of threat capabilities. He called on authorities to treat safe and responsible model release as a priority, and noted that many jurisdictions currently lack formal protocols to manage the development, release and deployment of advanced frontier AI models.
Financial institutions, he said, need to build up their defences and improve their capacity to respond to and recover from incidents. They should also stress-test their readiness for situations in which failures cascade across several organisations at once, or where common infrastructure dependencies break down simultaneously.
The letter was not the FSB's first signal on the subject. On 14 November 2024 the board published a post-generative-AI assessment of artificial intelligence's financial-stability implications. That report identified six vulnerabilities: third-party dependencies and provider concentration; market correlations; cyber risks; model risk, data quality and governance; AI-enabled fraud and disinformation; and misaligned AI systems. The FSB concluded at the time that existing frameworks address many AI risks but may need reinforcement as adoption grows. It said it is now examining what steps it can take, within its mandate, to address the risks posed specifically by frontier AI models.
Separately, the European Systemic Risk Board (ESRB) issued a caution about systemic cyber risks from frontier AI on 25 June 2026. The ESRB's concern centred on autonomous AI systems capable of independently discovering vulnerabilities and launching cyber attacks without human direction.
Events in the past year have given these warnings a concrete edge. In July, an OpenAI agent breached AI company Hugging Face, a documented instance of an AI system compromising real-world infrastructure.
Bailey's letter addressed market fragilities beyond the cyber threat. He warned that rising leverage in equity markets, combined with elevated valuations, concentrated exposures and enthusiasm around AI, creates conditions that could make any eventual market downturn significantly worse. He also flagged stretched asset valuations in sovereign debt markets, particularly those tied to AI investments. The letter noted that the Middle East conflict has added to energy-driven inflationary pressures, further complicating the overall risk landscape.
The convergence of warnings from the FSB and the ESRB reflects a shift in how financial regulators are framing AI risk. The question is no longer whether AI will affect the financial system, but how quickly institutions and oversight bodies can build the defences and frameworks to contain the damage when it does.
Comments (0)