US officials warn banks of Anthropic AI cyber risks
US officials warned bank CEOs this week about cyber risks from Anthropic's Mythos model. The AI can identify vulnerabilities in major operating systems.
In the United States, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an urgent briefing with the chief executives of major financial institutions this week. The high-level meeting was called to address significant cybersecurity risks associated with the latest artificial intelligence model developed by Anthropic.
The AI startup recently introduced a powerful model known as Mythos but has opted against a full public release. Anthropic indicated that the model possesses the capability to identify and exploit security flaws across all major operating systems and web browsers. Consequently, access to Mythos is currently restricted to approximately 40 technology partners, including MICROSOFT CORP and ALPHABET INC-CL A.
The discussion, held on Tuesday at the Treasury Department in Washington, focused on ensuring that the banking sector is prepared for the vulnerabilities exposed by Mythos and similar advanced AI systems. The meeting coincided with the presence of several bank leaders in the capital for other scheduled engagements.
Among the participants were the CEOs of CITIGROUP INC, MORGAN STANLEY, BANK OF AMERICA CORP, WELLS FARGO & CO, and GOLDMAN SACHS GROUP INC. While most major lenders were represented, JPMORGAN CHASE & CO chief Jamie Dimon was reportedly unable to attend the session.
While the Treasury Department and the Federal Reserve have remained tight-lipped regarding the specifics of the briefing, the initiative underscores growing government concern over the intersection of generative AI and national financial security. Both GOLDMAN SACHS GROUP INC and the Federal Reserve declined to comment on the proceedings, while other participants did not immediately respond to inquiries. This report includes contributions from journalists based in Mexico and New York.
