By The White House and Anthropic are in active discussions about deploying the AI firm’s powerful new model, Mythos Preview, within the federal government — even as ongoing efforts to blacklist the company as a supply chain risk continue, sources familiar with the discussions tell Axios.
Why It Matters
Anthropic is locked in a bitter dispute with the Pentagon over AI use in national security. Yet even U.S. officials who are critical of the company acknowledge that its AI tools could either strengthen American national security or pose serious risks if they fall into the wrong hands.
What’s Driving the Negotiations
Anthropic is rolling out Mythos Preview exclusively to a select group of companies and organizations, bypassing general public release. The controlled rollout is designed to help evaluate the model’s advanced cyber capabilities and allow organizations to harden their cybersecurity defenses before broader deployment.
- Several federal government agencies are seeking access to Mythos, and the White House and Anthropic are actively negotiating the terms of that access.
- Following inquiries from federal agencies about eligibility to use Mythos, the Office of Management and Budget sent an email—first reported by Bloomberg—confirming they were reviewing the matter.
- Two sources confirmed to Axios that discussions between Anthropic and government agencies are ongoing, with potential access expected within weeks.
The Pentagon Conflict: A Major Friction Point
These negotiations are unfolding despite Anthropic being embroiled in active litigation with the Pentagon, which has officially designated the AI company a “supply chain risk” and directed defense contractors to remove Anthropic software from military workflows.
- Anthropic is currently barred from Pentagon contracts but remains eligible to conduct business with other federal agencies while the legal case proceeds.
- “There’s progress with the White House. There’s not progress with [the Department of] War,” one administration official told Axios.
- A second administration official stated the government “has a responsibility to evaluate every model to see where the frontier of tech is,” while accusing Anthropic of deploying “fear tactics” by publicizing Mythos’s potential to enhance cyberattacks.
- “They’re using this Mythos cyber weapon to find friendly ears in the government,” the official said. “They’re succeeding.”
What Officials Are Saying
“All the intel agencies use Anthropic. Every agency except War wants to. That’s because Anthropic doesn’t want to kill people, and War’s position is ‘don’t tell us what to do.’ But if you’re the Department of Energy, you don’t care about that. You’re worried about the Chinese attacking the energy grid. So you want Anthropic,” the first administration official explained.
Key tensions in the debate include:
- Anthropic’s official policy prohibits its AI models from being used for mass surveillance or the development of fully autonomous weapons.
- The Pentagon argues that these restrictions are unduly limiting, citing the vague definitions of those terms and asserting it needs guarantees that AI systems can be used for “all lawful purposes.”
- These concerns are largely irrelevant to civilian agency use cases.
- Both Anthropic and the Pentagon declined to comment.
Civilian Agencies and Critical Infrastructure Defense
Federal civilian agencies such as the Departments of Energy and Treasury are tasked with safeguarding critical national infrastructure, including the U.S. electric grid and financial systems.
Gaining access to Mythos would enable these agencies to identify cybersecurity vulnerabilities within private companies and local governments and develop strategies to strengthen national cyber resilience ahead of potential attacks.
The Political Undercurrent
Senior officials within the Trump administration have characterized Anthropic’s leadership as ideologically “woke” and alarmist—and some reportedly welcomed the “supply chain risk” designation as a political statement.
However, those same officials and many others across the administration privately concede that Anthropic’s AI models are among the most capable available for national security applications.
One Pentagon official told Axios, even at the height of the feud: “The only reason the talks were ongoing is these guys are that good.”
1 Comment