Senior US financial regulators have summoned bank executives and encouraged them to test Anthropic's new Mythos artificial intelligence model to detect security vulnerabilities, according to a report by Bloomberg. The meeting this week involved Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell.
The move comes as Anthropic, the AI company behind the model, is engaged in a legal battle with the Trump administration. The Department of Defense has designated Anthropic a supply-chain risk after negotiations broke down over limits on how the US government could use the company's AI models.
Major Banks Already Testing
While JPMorgan Chase was named as an initial partner with access to Mythos, other major institutions including Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are also reportedly testing the model. Anthropic announced Mythos this week but stated it would limit initial access.
The company claims that despite not being specifically trained for cybersecurity, Mythos is "too good at finding security vulnerabilities." Some industry observers have suggested this claim could be hype or a strategic enterprise sales tactic.
Regulatory Concerns on Both Sides of the Atlantic
The push from US regulators occurs alongside reported discussions by UK financial authorities about the potential risks posed by the Mythos model. The dual-track approach highlights global regulatory interest in both the capabilities and dangers of advanced AI in critical sectors like finance.
The situation presents a complex landscape where US executive branch agencies are at odds, with the Treasury and Fed promoting the model's testing while the Defense Department maintains its risk designation against Anthropic.
Context and Next Steps
Anthropic's court battle with the Trump administration centres on the legal and operational implications of the supply-chain risk designation. The outcome could affect how other advanced AI models are procured and used by the US government and its regulated partners.
For the banking sector, the regulator-encouraged tests of Mythos will provide critical data on whether generative AI can reliably enhance financial cybersecurity infrastructure, potentially setting a precedent for future AI adoption in regulated industries.