Federal Reserve and Treasury Officials Discuss Cyber Risks
This week, Federal Reserve Chairman Jerome Powell and Treasury Secretary Scott Bessent convened a meeting with top executives from major U.S. banks. The topic was the potential cyber threats related to Anthropic’s Mythos model, as confirmed by CNBC.
Bank CEOs were already in Washington, D.C., for a Financial Services Forum board meeting when they were called for this special session on Tuesday. Sources familiar with the matter shared this information under the condition of anonymity.
Notably, Jamie Dimon, CEO of JPMorgan, was the only major bank leader who could not attend. Other prominent figures like Brian Moynihan from Bank of America, Jane Fraser from Citigroup, David Solomon of Goldman Sachs, Ted Pick from Morgan Stanley, and Charlie Scharf of Wells Fargo were all present, according to Bloomberg.
Earlier in the week, these CEOs had gathered for dinner, which likely contributed to the quick assembly for the Treasury Department meeting.
Bloomberg and the Financial Times were the first to report on the emergency meeting. As for comments, both the Federal Reserve and the Treasury declined to respond to inquiries, while Anthropic also has yet to comment.
This week, Anthropic made a limited preview of its new AI model, Claude Mythos, amid concerns that hackers could misuse its features.
JPMorgan Chase is one of the initial partners in a cybersecurity initiative called Project Glasswing, which also includes firms like Apple, Google, Microsoft, and Nvidia.
Officials at Anthropic mentioned that they are having “ongoing discussions” with the U.S. government, including relevant cybersecurity agencies, about the capabilities of the Claude Mythos Preview.
Recently, there was a drop in cyber stock prices after a draft blog from Anthropic came to light, detailing the advanced features and risks associated with the model.
The CEO commented on the potential implications, stating, “Getting this wrong clearly is dangerous, but if we get it right, we have a real opportunity to build a more secure internet.”
As part of Glasswing’s rollout, he had also noted the misuse of Anthropic’s model in past cyber attacks, specifically referencing an incident in November involving a Chinese group automating hacks against governments and corporations.
Moreover, Anthropic is currently engaged in a legal dispute with the Department of Defense concerning a supply chain risk identified as a national security concern. A federal appeals court recently denied the company’s request for a temporary block on its blacklisting.
