US Treasury secretary Scott Bessent called a meeting with the leaders of the US’s top banks on Tuesday to discuss the threat posed by Anthropic’s latest model, Bloomberg has reported.

The meeting was attended by executives from Bank of America, Citigroup, Goldman Sachs, Morgan Stanley and Wells Fargo, as well as Federal Reserve chair Jerome Powell, according to people familiar with the matter. JPMorgan Chase chief executive Jamie Dimon was also invited, but was unable to attend.

It was called in response to the launch of Anthropic’s latest AI model, Claude Mythos Preview. The software is designed to detect cybersecurity weaknesses and has been rolled out to a limited number of companies due to concerns over its use by bad actors.

Mythos Preview has already identified thousands of “zero-day” vulnerabilities across major systems and software. The Trump administration is concerned that these could be exploited by hackers, the Financial Times reported.

In his annual letter to shareholders, Dimon wrote that cyber risks remain “one of the biggest” threats to the financial industry, and that “AI will almost certainly” make that risk worse.

Mythos Preview is part of a wider cybersecurity initiative dubbed Project Glasswing, under which partners such as Amazon Web Services, Microsoft and Google will test their software using the model to find critical security flaws.

Anthropic is currently in discussions with the US government over the model’s capabilities and the related national security implications.

In January, the UK Treasury Committee called for widespread stress testing of AI usage in financial services amid concerns about fraud and unregulated financial advice.

---