FSB warns G20 that AI could trigger a market correction and sharpen cyberattacks on banks

The Financial Stability Board, which coordinates financial regulators for the G20, has told the group's finance ministers and central bankers that artificial intelligence poses two risks to the global financial system: a fall in AI-related valuations that spreads into a wider market correction, and increasingly sophisticated AI-enabled cyberattacks on critical banks and market infrastructure. FSB Chair Andrew Bailey, who is also Governor of the Bank of England, set them out in a letter published on 12 October ahead of the G20 meeting in Bangkok on 15 October, along with the FSB's first practical response: sound practices for how financial firms adopt AI.

What did the FSB warn the G20 about AI?

That AI is now one of the main risks to financial stability. Bailey's letter lists, alongside volatile sovereign bond markets and private credit, "risks associated with artificial intelligence (AI), including the potential for a market reassessment of AI-related valuations to trigger a broader market correction, as well as increasingly sophisticated AI-enabled cyber threats to the resilience of critical financial institutions and financial market infrastructures."

"AI therefore requires our continued close attention alongside careful management by financial institutions," he writes. Cointelegraph's account of the letter likewise leads on the warning about cyber threats enabled by AI.

Could an AI bubble trigger a market correction?

The FSB thinks it could spread one. The letter's concern is a "market reassessment of AI-related valuations," in which a repricing of AI-linked assets triggers "a broader market correction." The FSB is not predicting a fall; it is flagging how one would travel.

Bailey made the mechanism clearer in August. In that letter, the FSB said, he warned that "the recent increase in the use of leverage in equity markets is interacting with high valuations, market concentration and AI-related optimism in a way that could amplify a future market correction." Borrowed money, concentrated holdings and optimism about AI are what turn a repricing into a cascade.

Why does the FSB see AI cyberattacks as a threat to the financial system?

Because the systems are concentrated and the attackers are improving. The October letter warns of "increasingly sophisticated AI-enabled cyber threats" to critical institutions and market infrastructure, and says that "As AI adoption expands, financial institutions should address associated risks, including increasingly sophisticated cyber threats."

The August warning was sharper. Frontier AI models, the FSB said then, are "showing increasingly sophisticated autonomy and problem-solving abilities, as well as threat capabilities," and "The most immediate concern is the potential impact of frontier AI on cyber risk." Bailey asked jurisdictions to support "safe and responsible model release and deployment as a priority," and asked financial firms for robust response and recovery capabilities, including resilience among critical third-party technology providers.

That concern has a public record this year. OpenAI's agents chained software flaws to break into Hugging Face during an evaluation, and an agent got past the blocks on an Australian government portal, which we covered in our item on OpenAI's training pause. Neither involved a bank, and the FSB does not cite them; they show what autonomous systems with weakened safeguards have already done to real infrastructure.

What are the FSB's sound practices for responsible AI adoption?

A toolkit for financial firms, and the FSB's "first response on this topic," Bailey writes. The practices have been "designed as a toolkit, adaptable to the size, type, and level of AI adoption within financial institutions," and "span organisation-wide AI governance, lifecycle management, cyber risk mitigation, and management of third-party services." They incorporate feedback from a public consultation the FSB ran in June, and Bailey notes industry engagement helped deliver them at speed.

"To foster responsible adoption, I urge the G20 to champion these practices," he writes. He adds that there are "broader macroprudential issues that may warrant continued consideration and coordination," and the FSB says safe adoption of AI "will remain an important feature of the FSB's agenda next year."

Are the FSB's AI practices binding on banks?

No. The FSB coordinates national financial authorities and promotes policies; it does not make law. Its sound practices are guidance that matters as far as national supervisors and firms adopt them, which is why Bailey is asking the G20 to "champion" them rather than announcing a requirement. A bank in a G20 country is bound by what its own supervisor turns the practices into, if anything.

What happens at the G20 finance meeting in Bangkok on 15 October?

G20 finance ministers and central bank governors are due to consider the FSB's deliverables, which also include a consultation on modernising regulation and supervision, a review of public-sector backstop funding for banks in resolution, and a note on cross-border risks from global stablecoins. The reports were requested by the US G20 Presidency.

The meeting is the first test of whether the G20 takes up the AI practices. For AI companies, the more important line may be the one the FSB keeps repeating to governments: that safe release and deployment of frontier models is a financial-stability question, not only a technology one. That is the same question the US is answering this autumn with a voluntary industry pledge, which we covered in our item on the White House accord and the FTC probe.