Bank of England Governor warns G20 about threat of advanced AI models to financial system
Advanced artificial intelligence models could significantly increase the speed, scale, and cost-effectiveness of cyberattacks and potentially threaten the entire financial system. This was the warning from Andrew Bailey, chair of the Financial Stability Board (FSB) and Governor of the Bank of England, to G20 finance ministers and central bank governors.
Bailey's letter was publicly released on 31 August ahead of the G20 finance bloc's meeting in Asheville, North Carolina, taking place on 31 August and 1 September.
The FSB considers the potential impact of so-called frontier AI—the most powerful models at the leading edge of technology development—on cybersecurity to be the most immediate risk to the financial system.
Such systems are demonstrating increasingly sophisticated capabilities for autonomous operation, problem-solving, and vulnerability discovery. The FSB assesses that technological development could fundamentally alter the economics of cyberattacks: what previously required significant human resources, time, and specialized knowledge could become cheaper and faster.
Time between vulnerability discovery and attack is shrinking
Similar risks were also described in July by the Bank for International Settlements (BIS). Its researchers concluded that frontier AI increases the speed, scale, and sophistication of cyberattacks, although it could also be used for system defense.
The problem lies in the asymmetry: new tools may lower attackers' costs faster than financial institutions can rebuild their defenses. Frontier models are already capable of helping to find software vulnerabilities, and the time from discovery to potential exploitation in an attack could shorten dramatically.
For banks and other financial firms, this creates a particular challenge due to infrastructure interdependencies. A single critical vulnerability in widely used software or in a major technology provider's system could potentially affect many institutions at once.
Bailey called on regulators in different countries to coordinate rules for the safe release and deployment of the most powerful AI models. Financial firms, for their part, need to strengthen recovery capabilities after attacks and verify the resilience of external technology providers on which their critical services depend.
AI boom could also magnify future market downturns
Cyberthreats are not the only risk associated with the current development of AI. The FSB also drew attention to the situation in financial markets, where optimism over AI coexists with high asset valuations, capital concentration, and increasingly active use of leverage.
The FSB assesses that this combination could amplify the consequences of a future market correction. Should expectations for returns on AI investments deteriorate sharply, investors' use of borrowed funds could force them to sell assets more quickly, thereby increasing the scale of the decline.
The BIS previously warned about growing financial links between the AI sector and the private credit market. It estimates that over the past five years, direct lending funds have quadrupled the financing provided to companies in the AI and information technology sectors—roughly to around 15% of their portfolios.
Markets remain vulnerable to multiple simultaneous risks
The warning about AI came amid broader FSB concerns about the state of global markets. Bailey pointed to the risk of a potentially disorderly correction that, through global financial linkages, could rapidly spread across countries.
Among other vulnerabilities, the FSB cites high sovereign debt, troubles in private credit markets, and inflated valuations of some assets. Additional pressure comes from the Middle East conflict, which heightened inflation risks through rising energy prices.
The FSB stresses that AI itself holds significant potential to improve the efficiency of the financial sector and risk management. However, the pace of development of the most powerful models means that regulators and banks need to adapt cybersecurity frameworks even before malicious actors begin widely exploiting new AI capabilities.
Based on materials from: Financial Stability Board, Bank for International Settlements
