U.S. Treasury and Federal Reserve Address Cyber Risks of New AI Model with Bank CEOs

Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an urgent meeting with leaders of major U.S. banks this week to address cybersecurity risks associated with Anthropic’s new AI model, Mythos. The meeting was held at the U.S. Treasury headquarters in Washington on Tuesday ^[1][2].

The meeting was attended by the CEOs of Goldman Sachs, Bank of America, Citigroup, Morgan Stanley, and Wells Fargo. Notably, JPMorgan Chase CEO Jamie Dimon was invited but did not attend ^[1][2].

Anthropic’s Mythos model has been described as having the capability to identify and exploit vulnerabilities in major operating systems and web browsers. This has raised concerns among financial institutions and government officials about potential cybersecurity threats ^[3][4].

The release of Mythos has been limited to a select group of organizations, including Amazon, Apple, Microsoft, JPMorgan Chase, Cisco, Broadcom, and the Linux Foundation, as part of a controlled rollout ^[3][5].

Anthropic has been in discussions with U.S. government officials regarding the offensive and defensive cyber capabilities of the Mythos model ^[3][6].

What Is Known

The meeting was initiated by Bessent and Powell to discuss the potential cybersecurity threats posed by the Mythos AI model. It was attended by key figures from major U.S. banks, excluding JPMorgan’s Jamie Dimon. Mythos is capable of identifying vulnerabilities in widely used software systems ^[1][2][3].

What Remains Unclear

Details about specific vulnerabilities identified by Mythos and the exact nature of the discussions between Anthropic and U.S. government officials remain undisclosed. Additionally, it is unclear whether any immediate actions or policy changes were decided upon during the meeting ^[3][6].