Vice President JD Vance and Treasury Secretary Scott Bessent last week questioned leading tech CEOs about the security of artificial intelligence models and how to respond to cyber attacks before Anthropic released its new Mythos model, CNBC has learned.
The meeting occurred over the phone, according to two people familiar with the matter who asked not to be named because the event was private.
One of the people said that Anthropic’s Dario Amodei, xAI’s Elon Musk, Google‘s Sundar Pichai, OpenAI’s Sam Altman, Microsoft‘s Satya Nadella, CrowdStrike‘s George Kurtz and Palo Alto Networks‘ Nikesh Arora also participated.
The tech CEOs met to discuss the security posture of large language models and safe deployment, according to the person. Officials also discussed how to respond if models scale in favor of attackers, they added.
An Anthropic official declined to comment on the meeting, but told CNBC on Friday that the company has been in touch with senior officials about cybersecurity in recent weeks and has made itself available to support “the government’s own testing and evaluation of the technology.”
“Prior to any external release, Anthropic briefed senior officials across the U.S. government on Mythos Preview’s full capabilities, including both its offensive and defensive cyber applications,” the official said. “Bringing government into the loop early — on what the model can do, where the risks are, and how we’re managing them — was a priority from the start.”
OpenAI and CrowdStrike declined to comment on the meeting. The White House and the other companies have not responded to CNBC’s requests for comment.
Anthropic rolled out its Mythos AI model to a limited group of companies on Tuesday as it assesses ways to prevent hackers from using it. Apple, Google, Microsoft, Nvidia, Palo Alto Networks and CrowdStrike were among the initial launch partners.
Bessent and Federal Reserve Chair Jerome Powell this week called a surprise meeting with the heads of the biggest U.S. banks to address the potential threat of Mythos, signaling further concern from the Trump administration about advanced cyber tools.
It was also a sign that Anthropic is still part of the conversation about AI in the White House, as President Donald Trump seeks to strip the startup’s Claude platform from federal agencies.
Anthropic’s legal challenge to the Department of Defense supply chain risk designation is still playing out in court after two opposing rulings on opposite sides of the country.
A federal appeals court on Wednesday denied the company’s request to temporarily block the blacklisting, weeks after a federal judge in San Francisco granted Anthropic’s request for a preliminary injunction in another legal challenge.
With the opposing rulings, Anthropic remains blocked from DOD contracts but can keep working with other federal agencies while the cases play out.
— CNBC’s Kate Rooney contributed to this article.
