The head of the Financial Stability Board has warned G20 regulators that new artificial intelligence models are becoming a growing threat to the global financial system.

Andrew Bailey, who also serves as governor of the Bank of England, sent a letter on Monday to central bank governors and finance ministers across the Group of 20 economies. The warning follows a series of recent incidents in which new AI models from companies including OpenAI, Anthropic and Meta were used to hack other organizations over the internet.

Bailey said the risk is becoming more complex as frontier AI models demonstrate greater autonomy, stronger problem-solving abilities and more advanced offensive capabilities.

Regulators Fear Cybersecurity Gaps

One of the key concerns is that advanced AI models could identify previously unknown weaknesses in the cybersecurity defenses of financial institutions. Regulators also fear that once a fix is deployed, AI systems could adapt quickly and find a way around it.

The European Central Bank has already taken steps to address the issue. It has asked banks in the eurozone to submit a formal action plan dealing with the risks from new AI models by October 31.

G20 finance officials met on Monday in Asheville, North Carolina. The Financial Stability Board acts as the coordinating body for financial regulators across all G20 nations, making Bailey’s warning part of a broader effort to align responses before national rules diverge further.

Risks Do Not Stop at Borders

In his letter, Bailey said an AI-driven cyberattack on one country’s financial system could easily spread beyond its borders. Shared technology providers and financial infrastructure connect institutions around the world, meaning a breach in one place could create ripple effects elsewhere.

He also warned that differences in legal frameworks and cyber capabilities between countries could themselves become a weakness in the global system.

Bailey called on regulators to make the safe release of new AI models a top priority, saying many countries still lack the proper rules to manage them.

“Many jurisdictions do not have the protocols in place to manage the development, release, and deployment of advanced frontier AI models, heightening risks for the financial sector and beyond,” he wrote.

Banks and other financial firms were also told to prepare for worst-case scenarios, including simultaneous disruptions across multiple companies that share the same technology. Bailey said firms must be ready to rebuild computer systems from scratch after a severe attack. He described this as restoring systems from “bare metal,” meaning the ability to recover critical data and operations even after total system failure.

The warning comes as AI development continues to accelerate across the private sector, with new models being released by major technology companies at a rapid pace. Regulators are now working to build frameworks that can keep up with that speed, while also accounting for the possibility that the same tools improving productivity could be repurposed for cyberattacks.