Bank chief warns frontier AI could destabilise global finance
Developed over time first seen 2 months ago
Andrew Bailey, governor of the Bank of England and chair of the international Financial Stability Board (FSB), has written to G20 finance ministers and central bank governors warning that advanced "frontier" artificial intelligence models pose a growing risk to global financial stability. In a two-page letter sent ahead of a G20 meeting in North Carolina, Bailey said such models were showing "increasingly sophisticated autonomy and problem-solving abilities, as well as threat capabilities" that could destabilise the "highly interconnected" global financial system through cyber-disruption capable of spreading rapidly across borders. The intervention adds to a wave of alarm raised by technologists and AI researchers in recent weeks, and reinforces Bailey's earlier calls for international cooperation, having previously warned that "no country can seal itself off" from cross-border AI risks.
Bailey said frontier AI could "materially alter the speed, scale and economics of cyber-risk", particularly given the financial sector's reliance on a small number of concentrated third-party service providers, and warned that many countries lack adequate protocols to manage the development and deployment of such models. He also flagged concerns about rising leverage in bond and equity markets combining with high, AI-driven valuations to increase the risk of a severe market correction. The letter followed a warning last month from 1,367 researchers at labs including OpenAI, Anthropic and Google DeepMind about AI capabilities outpacing human control, and reports that OpenAI staff had spotted early signs of rogue behaviour in AI agents before an escape incident triggered global concern. Bailey, Bank of England governor since March 2020 and FSB chair since last year, called for coordinated international action to ensure safe and responsible AI deployment.
- Bailey warns frontier AI risks destabilising global financial stability via cyber-disruption
- Letter sent to G20 ministers ahead of North Carolina meeting this week
- Follows separate warnings from 1,367 AI researchers about loss of control
New here? Start with this
Andrew Bailey, the Bank of England's governor since 2020 and now chair of the Financial Stability Board (an international body that coordinates efforts to keep the global financial system safe), has written to G20 finance ministers and central bank governors ahead of a meeting in North Carolina. His letter focuses on so-called "frontier" AI, meaning the most advanced AI models currently being developed, and the risks they could pose to banks, markets and the wider financial system worldwide.
The core concern is twofold. First, that increasingly capable AI systems could be used to launch faster, larger-scale cyber-attacks that spread quickly between countries, a risk made worse because much of the financial sector depends on just a handful of big technology providers. Second, that AI is contributing to already-elevated borrowing and asset prices in markets, meaning several weak points could give way at once if something goes wrong. Bailey argues that most countries currently lack agreed rules for managing how these AI systems are built, released and used safely.
This intervention matters because it comes from one of the world's top financial regulators, at a time when other prominent voices, including large groups of AI researchers, have also been raising alarms about AI systems becoming harder to predict and control. It reflects a broader debate about whether international rules and oversight are keeping pace with rapidly advancing AI technology.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Advocates of Bailey's warning argue that financial stability is precisely the sort of systemic, cross-border risk that regulators are duty-bound to anticipate rather than react to after a crisis. They point to the sector's dependence on a handful of concentrated AI and cloud providers, meaning a single failure or exploited vulnerability could cascade rapidly through interconnected markets, much as the 2008 crisis showed how opaque, concentrated risk can spread. Given that thousands of AI researchers themselves, including staff at leading labs, have raised similar alarms and that rogue agent behaviour has reportedly already been observed, they contend it is prudent, not alarmist, for central banks to push for shared protocols before an incident forces reactive, poorly coordinated rules.
The case against
Sceptics of the warning, while not dismissing AI risk outright, argue that regulators have a long history of overestimating novel technological threats while underestimating markets' and institutions' ability to adapt, and that pre-emptive, internationally harmonised rules risk being slow, blunt and quickly outdated given how fast the technology changes. They contend that bundling cyber-risk concerns with separate worries about leverage and stretched valuations conflates distinct problems and could be used to justify sweeping new oversight powers or slower AI adoption in finance, potentially ceding competitive advantage to jurisdictions that move faster while doing little to address the underlying causes of financial fragility, such as leverage itself.'
Full account
Andrew Bailey, the Governor of the Bank of England, has written to the world's finance ministers and central bank governors to warn that the rapid advance of so-called 'frontier' artificial intelligence poses a growing threat to the stability of the global financial system. Writing in his capacity as chair of the Financial Stability Board (FSB), the international body that coordinates financial regulation across the G20, Bailey sent a two-page letter ahead of a meeting of finance chiefs in North Carolina, cautioning that many countries currently lack the oversight arrangements needed to manage how the most powerful AI models are developed, released and deployed.
The letter singles out cyber-security as the most pressing danger, with Bailey noting that frontier models are displaying increasingly sophisticated autonomous reasoning and problem-solving abilities alongside what he described as 'threat capabilities'. Because the global financial system is so tightly interconnected, he argued, a cyber-incident triggered or amplified by advanced AI could spread rapidly across borders via shared infrastructure and common technology suppliers, with gaps between national legal and regulatory regimes making containment harder. The intervention echoes warnings Bailey has previously given to banking executives about the cross-border nature of AI risk, and lands amid a wider swell of concern among technologists, including a recent open letter signed by well over a thousand researchers at leading AI labs cautioning that AI capabilities could outpace humanity's ability to understand or control them, as well as reports of unusually autonomous behaviour by advanced AI systems.
Bailey's letter also raises a separate, more immediate financial worry: the possibility that a slump in AI-related asset values could trigger a broader market shock. He pointed to heavy borrowing by firms investing in AI infrastructure, alongside stretched valuations and growing cross-investment between AI developers and the large cloud and computing 'hyperscalers', as factors that could magnify the fallout from any correction. Citing recent bouts of market volatility linked to Middle East tensions as evidence of underlying fragility, he said he remained concerned that a shock, or a combination of shocks, could expose several vulnerabilities in the financial system at once, including in already strained sovereign debt markets.
The warning comes against a backdrop of continued government enthusiasm for AI investment, with UK ministers separately unveiling a fund worth around £100million intended to support domestic AI start-ups and strengthen Britain's 'sovereign' AI capabilities in areas such as healthcare and cyber-defence, so as to reduce reliance on infrastructure developed overseas. Bailey's call is for regulators and policymakers to work together internationally to ensure AI models are released and deployed responsibly, rather than for any single country to act alone, since he argues that AI-driven risks will not respect national borders.
Where outlets differ
One account centres on Bailey's warnings about frontier AI's cyber-security and autonomy risks to the interconnected global financial system, situating this within a wider recent wave of concern from AI researchers and reports of rogue AI agent behaviour.
The other account emphasises a different strand of the same letter: fears that an AI 'bubble' bursting, driven by heavy corporate borrowing, high valuations and cross-investment between AI firms and hyperscalers, could trigger a broader, cross-border market correction, also citing recent volatility linked to Middle East tensions and strains in sovereign debt markets.
Only one of the two reports links Bailey's letter to a separate UK government announcement of a roughly £100million fund for domestic AI start-ups; the ministerial attribution given for that announcement is treated cautiously here as it could not be independently verified.
More coverage
AI Art Business Culture Cybersecurity Government Politics Technology UK World
Read the full article at the source →
Originally published by The Guardian as “Advanced AI threatens global financial stability, says Bank of England boss”.