Federal Reserve Chairman Jerome Powell and Treasury Secretary Scott Bessent met with CEOs of major U.S. banks this week to discuss potential cyber risks posed by Anthropic’s Mythos model, CNBC confirmed Friday.
Bank leaders were already in Washington, D.C., to attend a Financial Services Forum board meeting when a special meeting was called to discuss the Mythos on Tuesday, according to people familiar with the matter, who requested anonymity to share confidential information.
The CEOs were summoned to a dinner by the Treasury Department earlier in the week, one of the people said. JP Morgan”Jamie Dimon was the only major bank CEO who was unable to attend the meeting, they added.
bank of americabrian moynihan citygroupJane Fraser, goldman sachs CEO David Solomon said: morgan stanleywith Ted Pick of wells fargo CEO Charlie Scharf and others all attended, Bloomberg reported, citing people familiar with the meeting.
Bloomberg and the Financial Times first reported the meeting with bank executives.
The Federal Reserve declined to comment to CNBC. The Treasury Department did not respond to requests for comment.
The surprise meeting between top bankers and two of the most powerful federal financial regulators signaled that advanced AI capabilities are a top concern for the Trump administration and could threaten the foundations of the U.S. financial system.
Earlier this week, Anthropic released a limited preview of its new artificial intelligence model, Claude Mythos, due to concerns that hackers could exploit its features.
Banking giant JPMorgan Chase was one of the first launch partners of the cybersecurity initiative known as Project Glasswing. Other partners include: apple, google, microsoft and Nvidia.
Ahead of its release, the company briefed U.S. government officials on the model and its “offensive and defensive cyber applications.”
Anthropic officials told CNBC that the company is in “ongoing discussions” with the U.S. government, including the Cybersecurity and Infrastructure Security Agency and the Center for AI Standards and Innovation, about the cyber capabilities of Claude Mythos Preview.
“It’s clear that getting this wrong is dangerous, but if we get it right, we have a real opportunity to build an Internet and world that is fundamentally more secure than it was before the advent of AI-powered cyber capabilities,” CEO Dario Amodei said in a post on X accompanying Glasswing’s rollout.
The Trump administration’s involvement with the Mythos model comes amid Anthropic’s recent challenge to the Department of Defense over the company’s designation of its AI lab as a supply chain risk to national security.
President Donald Trump and Secretary of Defense Pete Hegseth publicly criticized the company for advocating limits on the use of AI technology in warfare, and the president ordered federal authorities to stop using Anthropic’s platform.
CNBC previously reported that the Pentagon continued to use Claude during the Iran war.
A federal appeals court this week denied the company’s request to temporarily block its blacklisting. In March, a federal judge in San Francisco granted Anthropic’s request for a preliminary injunction in a separate lawsuit.
The duel ruling means Anthropic remains barred from contracts with the Department of Defense but can continue to work with other government agencies while their respective legal challenges proceed.
Cyber stocks fell late last month after Fortune uncovered a draft blog post from Anthropic that revealed the model’s advanced cyber features and potential risks.
Hackers have used Anthropic’s model in the past to orchestrate AI-powered attacks.
In November, the company revealed that a Chinese group used Claude to automate hacks targeting governments and businesses.
