FSB Chair Andrew Bailey warns that increasingly autonomous frontier AI could amplify cyber threats across a financial system already exposed to concentrated technology providers #
The growing autonomy of 'frontier' AI models could transform the speed, scale and cost of cyber attacks on the global financial system, the head of the world's top financial stability watchdog has warned G20 ministers and central bankers.
In a letter to counterparts ahead of key meetings in Asheville, North Carolina, Financial Stability Board Chair Andrew Bailey said advanced systems were showing increasingly sophisticated autonomy, problem‑solving abilities and threat capabilities, with cyber risk now his most immediate stability concern.
This warning, dated 28 August, was addressed to G20 finance ministers and central bank governors. Bailey highlighted that banks and other financial firms rely heavily on a relatively small number of technology providers, which he suggested could magnify the impact of any AI-enabled attack.
Frontier AI Seen as a Potential Game-Changer for Cyber Crime #
Bailey said more capable models could alter how quickly and cheaply cyber threats are developed and deployed. That could increase the potential scale of attacks against financial institutions and critical infrastructure.
The concern extends beyond a single bank being compromised. A major technology provider used by multiple financial firms could become a common point of failure, allowing disruption to spread through shared systems and infrastructure.
The FSB said these risks would not respect national borders. Common technology providers, cloud infrastructure and cross-border financial activity could allow an incident originating in one jurisdiction to affect firms elsewhere.
The warning comes as some AI companies disclose incidents involving models attempting to bypass safeguards or behave outside their intended parameters.
Those episodes provide a backdrop to Bailey's concern about increasingly autonomous systems, although the FSB letter does not identify individual incidents as evidence of an imminent financial crisis.
Heavy Reliance on Tech Giants Raises Systemic Questions #
The FSB has previously warned about financial institutions' dependence on critical third-party technology providers. Bailey said that concentration could become more consequential as frontier AI becomes more capable.
The letter does not identify OpenAI or claim that 69 per cent of financial institutions depend on the company for core operations.
That figure, which has circulated alongside the FSB warning, comes from separate industry research and should not be attributed to the watchdog.
Instead, the FSB describes the concentration problem in broader terms. If many firms depend on the same provider, an attack or serious technical failure could undermine market confidence across institutions at the same time.
Bailey was not suggesting that an AI-driven liquidity crisis is already developing. He was warning about how the characteristics of frontier models could amplify an existing operational and cyber-risk problem.
Regulators Scramble To Catch up With Frontier Models #
Bailey also pointed to a regulatory gap. Many jurisdictions, he said, do not yet have adequate protocols governing the development, release and deployment of advanced frontier models.
The FSB is calling for steps to support the safe and responsible release and deployment of such systems globally. It also stressed the need for financial firms to maintain strong response and recovery capabilities as cyber threats become more capable.
One recommendation has an intentionally old-fashioned sound: firms should be able to recover critical systems and data from 'bare metal' after a major cyber incident.
The concept refers to restoring systems from a clean, trusted environment rather than relying on infrastructure that may have been compromised. Reporting by the Financial Times said Bailey had urged firms to maintain backups isolated from their main networks.
It is described as a resilience measure, not a proposal to permanently disconnect financial markets from the internet.
AI Risk Runs Through Cyber Security and Markets #
Bailey's warning also sits within a wider FSB assessment of financial stability. The board has flagged leverage, high valuations, market concentration and optimism around AI as factors that could amplify the effects of a disorderly market correction. The AI concern therefore runs through two related channels: cyber risk and market vulnerability.
The FSB is not saying autonomous AI has already escaped financial controls or that models are independently moving markets. Its concern is that increasingly capable systems could make cyber attacks faster, larger and potentially more difficult to contain, while financial institutions remain connected through the same technology providers.
© Copyright IBTimes 2026. All rights reserved.