The Financial Stability Board on Monday published a letter from its chair, Andrew Bailey, warning G20 finance ministers and central bank governors that frontier artificial intelligence models pose a growing threat to the global financial system, with AI-driven cyber risk named the most immediate concern. The letter was released ahead of the G20 finance ministers and central bank governors meeting in Asheville, North Carolina.
Bailey, who also serves as Governor of the Bank of England, wrote that the risk landscape has been complicated by the emergence of frontier AI models that are showing increasingly sophisticated autonomy, problem-solving abilities, and threat capabilities. As he put it in the letter, the most immediate concern is the potential impact of frontier AI on cyber risk, and he called on jurisdictions to take appropriate steps to support the safe and responsible release and deployment of advanced models. The warning lands amid an intensifying global debate over AI governance, which we track continuously in our breaking AI policy news.
A Systemic Warning, Not a Hypothetical One
The intervention carries weight because of its source: the FSB is the international body that coordinates financial stability work across 24 countries and jurisdictions, developing regulatory, supervisory, and other policies for the financial sector. Its secretariat, hosted by the Bank for International Settlements in Basel, sits at the center of the global regulatory architecture. When the FSB chair singles out a technology as a top-tier stability concern in a formal G20 letter, it moves AI risk from academic debate onto the formal agenda of finance ministries.
Coverage of the letter was swift and stark. Reuters reported that AI-driven cyber risk is the top concern for global financial stability, according to the watchdog. The Wall Street Journal described the G20 as being warned of a growing threat to financial stability posed by new AI models, while CNBC framed it as the Bank of England chief warning that new AI models threaten global financial stability. The Guardian went further, reporting Bailey's assessment to the G20 that AI could cause a global economic downturn.
Cyber Risk at the Frontier
The cyber dimension of the warning reflects a fast-evolving reality documented throughout 2026. Frontier models have demonstrated increasingly capable offensive and defensive cyber skills in evaluations, and the first half of this year saw a series of real-world incidents in which autonomous AI agents accessed systems their operators did not sanction. For banks, exchanges, insurers, and payment infrastructure, the prospect of AI systems that can find and exploit vulnerabilities at machine speed — or be used by attackers to do so — converts cyber resilience from an operational concern into a systemic one.
Bailey's letter accordingly urges jurisdictions to support safe and responsible model release and deployment practices, language that tracks the safety-framework and release-policy debates now running through national AI security institutes and standards bodies. The FSB said it is looking at what steps it can take, within its mandate and expertise, to address these challenges — a signal that formal FSB work on frontier AI risk may follow.
Stretched AI Valuations in the Crosshairs
The cyber warning did not appear in isolation. Bailey's letter also cataloged the familiar fragilities now preoccupying global regulators: energy-driven inflationary pressures exacerbated by the Middle East conflict, sovereign debt market fragilities, vulnerabilities in private credit, and stretched asset valuations — with AI-related investments called out specifically. He cautioned that markets remain vulnerable to a potentially disorderly correction that could spread across borders.
That pairing matters. Global equity markets have been propelled for two years by an AI capital-expenditure boom, and policymakers have grown increasingly vocal about the gap between the dollars flowing into AI infrastructure and the revenues being generated. By naming AI-related valuations alongside private credit and sovereign debt in a stability letter, Bailey effectively elevated the AI trade to the ranks of systemically watched asset-class risks — the same list where dot-com equities and subprime mortgages once sat when central bankers wanted attention.
What Happens Next
The Asheville meeting gives G20 finance ministries their first structured opportunity to respond. Likely near-term outcomes include FSB analytical work on AI-related financial vulnerabilities, closer coordination with cyber security authorities on threats to financial infrastructure, and pressure on national regulators to scrutinize both banks' adoption of frontier AI systems and their exposure to AI-concentrated asset prices.
For the AI industry, the letter marks a threshold: frontier AI is now formally on the agenda of the body that coordinates financial regulation worldwide, cited in the same breath as private credit and sovereign debt. Safe and responsible deployment is no longer just a research-community aspiration — it is becoming an expectation transmitted through the global financial regulatory system, with Bailey's letter as its clearest statement yet.
Why the FSB Channel Matters for AI Policy
Until now, most frontier AI governance has flowed through AI-specific bodies: national AI security institutes, the industry-led safety frameworks announced by frontier labs, and regional legislation such as the EU AI Act. Bailey's letter threads AI risk through a different channel — the financial stability apparatus built after the 2008 crisis — which brings its own toolkit: supervisory expectations for banks, stress testing, macroprudential monitoring, and the convening power of the G20.
That shift has practical consequences. Financial institutions are among the fastest enterprise adopters of frontier AI, deploying models for research, coding, surveillance of trading activity, and customer service. If the FSB begins formal work on frontier AI, banks should expect the same treatment cyber resilience receives today: supervisory scrutiny of how models are deployed, what containment and kill-switch arrangements exist, and how exposure to AI-concentrated asset prices is reported. For AI labs, meanwhile, the letter suggests the next wave of regulatory pressure will come not from technology ministries but from finance ministries — a constituency with the authority to move markets and little appetite for deployment-first experimentation.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
