Dario Amodei Calls OpenAI Pentagon Deal Claims ‘Lies’

A war of words has erupted between two of the world’s leading companies. Anthropic CEO Dario Amodei has publicly accused OpenAI and its CEO Sam Altman of dishonesty over the Pentagon’s AI deal. He sharply criticized messaging from Sam Altman and OpenAI regarding a military agreement with the U.S. government.

Last week, Anthropic and the U.S. Department of Defense failed to reach a particular agreement. The military intended to have open access to Anthropic’s AI technology. Before making this new deal, Anthropic already had a $200 million contract with the DoD, which is why the company stated it would remain neutral because of safety concerns.

Anthropic requested that DoD confirm it would not use its AI for domestic mass surveillance or autonomous weaponry. The DoD did not provide clear assurances and just stated that the technology will be used for “any lawful use.” Because of this lack of clarity, the negotiations collapsed.

Within the hours of this failed agreement, the Pentagon turned to OpenAI. Sam Altman quickly announced a new defense contract with the Department of Defense. He stated the deal included protections against the same red lines Anthropic had drawn. OpenAI’s blog post claimed mass domestic surveillance was explicitly excluded under “lawful use.”

However, Amodei was unconvinced and rejected those claims in an internal message. In an internal memo, Amodei described OpenAI’s public framing of its Pentagon contract as misleading, and OpenAI’s messaging about the deal as “straight up lies.” He accused Altman of falsely “presenting himself as a peacemaker and dealmaker.” He labelled the entire arrangement “safety theater.”

“The main reason [OpenAI] accepted [the DoD’s deal] and we did not is that they cared about placating employees, and we actually cared about preventing abuses,” Amodei wrote to Anthropic staff.

Critics have noted that “lawful use” is a shifting standard. What is considered illegal today could be legally permitted tomorrow. That ambiguity sits at the heart of Anthropic’s objection to OpenAI’s deal.

Amodei did not stop at policy criticism. He turned his attention to public perception and wrote in the memo, “I think this attempted spin/gaslighting is not working very well on the general public or the media, where people mostly see OpenAI’s deal with the DoW as sketchy or suspicious, and see us as the heroes (we’re #2 in the App Store now!).”

He continued, “It is working on some Twitter morons, which doesn’t matter, but my main worry is how to make sure it doesn’t work on OpenAI employees.”

He added that his primary concern was protecting OpenAI’s own employees from the spin. The memo was a rare, unfiltered look inside the rivalry between the two companies.

The controversy showcases the increasing issues in the AI industry as companies navigate government partnerships and national security demands. Developers are day-by-day debating where to keep controls when powerful models are integrated in the military environments.

More broadly, the episode highlights how AI leaders are struggling to define ethical boundaries for powerful models as government interest in the technology rapidly expands.