Anthropic says talks with the Pentagon have barely moved forward, and the company still can’t accept what defense officials are calling their final offer on AI safeguards. CEO Dario Amodei made the comment Thursday as a Friday afternoon deadline looms, 5:01 p.m. to be exact, for Anthropic to either let the Department of Defense use its Claude model without the restrictions it wants or face serious fallout.
In a statement, Anthropic said: “The contract language we received overnight from the Department of War made virtually no progress on preventing Claude’s use for mass surveillance of Americans or in fully autonomous weapons.” They added that what was presented as a compromise came with wording that would let those safeguards be ignored whenever the Pentagon felt like it. “Despite DOW’s recent public statements, these narrow safeguards have been the crux of our negotiations for months.”
The company made clear it’s not walking away from the table. Negotiations are expected to keep going right up to the deadline. The Pentagon hasn’t commented publicly on Anthropic’s latest statement.
Why This Fight Matters
The core disagreement is simple: Anthropic wants firm rules that Claude can’t be used for mass surveillance of U.S. citizens or for fully autonomous weapons. The Pentagon insists its AI models must be available for “all lawful purposes” in classified settings, no exceptions.
That “all lawful purposes” standard isn’t new or unique to Anthropic. It’s the same requirement the department applies across the board. xAI already signed a contract under those terms for classified work. Talks to bring OpenAI and Google into classified use are moving fast too, as per a report from Axios. Anthropic has been the only company so far whose model has actually been used in classified environments.
What Could Happen Next
If no deal is reached by Friday evening, the Pentagon has already started laying the groundwork for consequences. Defense contractors such as Boeing and Lockheed Martin were recently asked to review their exposure to Anthropic, code for assessing whether they could be hurt if the company gets blacklisted as a supply chain risk.
Pete Hegseth, the Defense Secretary, has also threatened to use the Defense Production Act to force Anthropic to hand over Claude without any limits. Legal experts say that kind of order would sit on shaky ground.
Amodei said Anthropic remains committed to engaging with Pentagon on the issue. “It is the Department’s prerogative to select contractors most aligned with their vision. But given the substantial value that Anthropic’s technology provides to our armed forces, we hope they reconsider. Our strong preference is to continue to serve the Department and our warfighters—with our two requested safeguards in place,” the statement reads. The company still hopes to find a path forward before the clock runs out.
