Trump officials reportedly urging banks to explore Anthropic’s Mythos AI model

Officials from the Trump administration are reportedly encouraging major financial institutions to explore the capabilities of Anthropic’s newly introduced Mythos model, according to a Bloomberg report.

The report states that U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened a meeting with banking executives earlier this week. During the discussion, they reportedly encouraged participants to consider using Mythos to identify potential security vulnerabilities.

Although JPMorgan Chase was the only institution officially named among the initial partners granted access to the model, other major banks — including Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley — are also reportedly testing the system.

Anthropic unveiled Mythos earlier this week but indicated that access would remain limited for the time being. The company cited concerns about the model’s ability to uncover software vulnerabilities, noting that although it was not specifically trained for cybersecurity, it has demonstrated high effectiveness in that area. At the same time, some observers have suggested that these claims could be overstated or part of a broader enterprise positioning strategy.

The situation is particularly notable given Anthropic’s ongoing legal dispute with the Trump administration. The company is currently challenging a Department of Defence decision that classified Anthropic as a supply chain risk. That designation reportedly followed unsuccessful negotiations related to the company’s attempts to impose restrictions on how government agencies could use its AI models.

Separately, the Financial Times has reported that financial regulators in the United Kingdom are also evaluating the potential risks associated with the Mythos model, highlighting growing international attention around its capabilities and implications.