STORY: U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell held an urgent meeting with major bank CEOs this week, warning of cybersecurity risks linked to Anthropic's latest AI model.
That's according to two sources familiar with the matter.
Anthropic launched its powerful new "Mythos" model earlier this week but stopped short of a broad release, citing concerns it could expose previously unknown cybersecurity vulnerabilities.
It said Mythos can identify and exploit weaknesses [beat] across "every major operating system and web browser."
Last week, Anthropic said it was in ongoing talks with U.S. government officials about the "offensive and defensive cyber capabilities" of Mythos.
Anna Rathbun is founder and CEO of Grenadilla Advisory.
"AI has gotten so smart that it can actually wreak havoc on global systems like banking. [CUT] This isn't you and me using AI and, you know, AI hallucinating and then us correcting it. That is not where this is happening. This is happening at a very sophisticated enterprise level. So that worries me a bit. But what actually makes me happier, or less worried, is the fact that the creators of these AI platforms are actually not distributing it yet. And being worried themselves are actually coming to the regulators and coming to the banking leaders and asking them to have a discussion."
Anthropic says access to Mythos will be limited to about 40 technology companies, including Microsoft and Google.
One source said the meeting with Bessent and Powell was aimed at ensuring banks understand the risks posed by Mythos and similar models, and are taking steps to protect their systems.
Bloomberg News reported that the CEOs of Citigroup, Morgan Stanley, Bank of America, Wells Fargo and Goldman Sachs attended. JPMorgan CEO Jamie Dimon was unable to join.
Goldman Sachs, Wells Fargo and the Federal Reserve declined to comment. The Treasury, other banks and Anthropic did not immediately respond to requests for comment.
