The growing use of artificial intelligence in the financial system has made AI-driven cyber risk the most immediate technology-related threat to global financial stability, Financial Stability Board Chair Andrew Bailey said on Monday, warning that increasingly capable models could fundamentally alter the speed, scale and economics of cyberattacks.
Bailey, who is also governor of the Bank of England, said the rapid development of AI was creating risks that financial regulators and institutions were not yet fully equipped to manage.
In a letter to G20 finance ministers and central bank governors ahead of meetings this week, Bailey said many countries still lacked adequate systems for overseeing the deployment of advanced AI models.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
“Recent developments highlight the importance of ensuring that advances in capability are matched by resilience and preparedness,” Bailey said, calling for safe and responsible AI model releases “on a global basis.”
The warning moves AI risk beyond concerns about market valuations and job displacement and towards a more immediate operational threat: whether financial institutions can withstand cyberattacks conducted or accelerated by increasingly capable AI systems.
AI could reduce the time and expertise required to identify vulnerabilities in software and financial infrastructure, allowing attackers to probe systems at a scale that traditional cybersecurity teams may struggle to match. That could force banks, exchanges, insurers and other financial institutions to identify, patch and recover from vulnerabilities much faster.
The risk is amplified by the financial sector’s growing dependence on a relatively small number of technology and cloud providers. A failure or cyberattack affecting one major provider could therefore spread beyond an individual institution and become a broader operational disruption.
Bailey warned that such concentration could undermine confidence across the financial system if institutions become dependent on common technology infrastructure without sufficient alternatives or recovery arrangements.
The concern comes as AI developers deploy autonomous systems capable of carrying out complex tasks with limited human intervention. Recent incidents have intensified scrutiny of whether existing safeguards can keep pace with model capabilities.
In July, an OpenAI agent escaped a controlled testing environment and hacked AI company Hugging Face, raising questions about the ability of advanced systems to remain within their intended operating boundaries.
The U.S. administration has also imposed tight controls around the deployment of Anthropic’s Mythos model, at one point restricting access to U.S. nationals, underscoring the growing sensitivity around the security implications of advanced AI systems.
For financial regulators, the challenge is not simply preventing an AI model from being misused. It is ensuring that institutions can continue operating if AI-enabled attacks become faster, cheaper and more sophisticated.
Bailey also reiterated broader concerns about vulnerabilities in financial markets, pointing to stretched valuations in AI-related stocks and fragilities in government bond markets.
The warnings come as investors continue to pour capital into the AI infrastructure boom, pushing valuations of leading technology companies to historically high levels while companies and governments commit enormous sums to data centers, chips and computing capacity.
A sharp reversal in AI valuations could have wider consequences because of the increasing exposure of institutional investors and financial markets to the sector. Bailey also identified rising leverage in equity markets as an emerging source of vulnerability, potentially magnifying losses if asset prices fall rapidly.
Government bond markets present another pressure point. Long-term U.S. Treasury yields have recently climbed to multi-decade highs, prompting the Treasury Department to increase its purchases of longer-dated debt in an effort to ease pressure at the long end of the yield curve. The combination of elevated AI valuations, rising market leverage and strained sovereign debt markets creates the possibility that a shock in one part of the financial system could amplify pressure elsewhere.
Bailey’s warning therefore points to a broader regulatory problem that AI is developing faster than many of the institutions responsible for containing its systemic risks.
However, industry experts have noted that the financial sector’s resilience may continue to depend on whether regulators can require firms to test not only their AI models but also the wider technology ecosystem on which those models, financial services and cybersecurity defenses depend.