Anthropic Discussed New AI Model With Federal Government, Co-Founder Says

Anthropic has been in talks with the Trump administration about its new AI model, Claude Mythos Preview, despite the Pentagon designating it as a supply-chain risk, co-founder Jack Clark said on April 13.

The designation, which prevents Anthropic from doing business with the government and its contractors, followed the company’s refusal to grant the Pentagon unrestricted access to its Claude models due to concerns they could be used for mass domestic surveillance or fully autonomous weapons. The Pentagon has said it does not intend to use Claude for such purposes.

Clark said on Monday that Anthropic wants to keep the government informed about its new artificial intelligence (AI) models despite what he called “a narrow contracting dispute” with the Pentagon, citing the company’s commitment to national security.

“Our position is the government has to know ​about this stuff,” he said at the Semafor World Economy event in Washington. “So absolutely, we talked to them about ​Mythos, and we’ll talk to them about the next models as well.”

Clark did not elaborate on the details of Anthropic’s talks with the Trump administration.

The Epoch Times has reached out to the Department of War for comment but did not receive a response by the time of publication.

Anthropic announced Claude Mythos Preview on April 7, as part of its cybersecurity initiative, Project Glasswing. It said the model has not been released for the general public due to its “large increase in capabilities,” and is instead being shared with a limited group of technology and cybersecurity partners, including companies like Microsoft, Nvidia, Google, Apple, and Cisco.

Anthropic described the new AI model as its “most capable frontier model” that can help to detect and fix vulnerabilities within operating systems and web browsers. It can also be used to design “sophisticated ways to exploit those vulnerabilities,” according to the company’s report.

“Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. The fallout—for economies, public safety, and national security—could be severe,” it stated in a blog post.

Meanwhile, a federal appeals court in Washington ruled on April 8 that the Department of War may designate Anthropic as a supply-chain risk after the company asked the court to block the designation.

The appeals court’s three-judge panel determined that it would not block the designation because Anthropic “has not satisfied the stringent requirements for a stay pending court review.”

The Pentagon has previously used the Claude AI system, made by Anthropic, for mission-critical functions, including intelligence analysis, modeling and simulation, operational planning, and cyber operations.

Matthew Vadum contributed to this report.