The Pentagon clashed with rival contractor Anthropic over ethics limitations on its tech
RT composite. United States Secretary of War Pete Hegseth speaks during a visit to Sierra Space in Louisville, Colorado, February 23, 2026. © Getty Images / AAron Ontiveroz / The Denver Post; Samuel Boivin / NurPhoto
The
US Department of War has reportedly signed an agreement with Elon
Musk’s xAI to integrate its Grok chatbot into classified military
systems, escalating pressure on rival contractor Anthropic as it refuses
to lift safeguards on its Claude model.
The deal, first reported
by the New York Times and confirmed by Axios on Monday, would make Grok
the second AI system approved for use on the military’s most sensitive
networks, where intelligence analysis, weapons development, and
battlefield operations take place. Until now, Anthropic’s Claude has
been the sole model available on classified platforms, through a
partnership with Palantir Technologies.
The agreement comes as
Secretary of War Pete Hegseth summoned Anthropic CEO Dario Amodei for
what sources expect to be a tense meeting at the Pentagon on Tuesday.
According to Axios, Hegseth is expected to present an ultimatum: agree
to make Claude available for “all lawful purposes” without additional safeguards, or face consequences including potential designation as a “supply chain risk” –a label typically reserved for entities linked to foreign adversaries.
Anthropic has resisted Pentagon demands to remove restrictions that
prevent its technology from being used for mass surveillance of
Americans or deployed in fully autonomous weapons systems with no humans
in the loop.
xAI has reportedly agreed to the demands, but the company has yet to comment on the reports. Google is also reportedly “close” to a deal allowing classified use of its Gemini model, according to people briefed on the discussions, while OpenAI remains “not close” as it continues working on safety technology.
Pentagon officials acknowledge that replacing Anthropic within its
classified systems could cause short-term disruptions. The model was
used during the operation to capture Venezuelan President Nicolas Maduro
last month – the first known instance of AI playing a direct role in an
active military raid.
Anthropic has positioned itself as the
safety-conscious alternative within the AI industry. CEO Amodei has
repeatedly warned of the existential dangers posed by unconstrained
artificial intelligence, including “autonomy risks.” The company’s Safeguards Research Team lead, Mrinank Sharma, abruptly resigned last week with a cryptic warning that “the world is in peril.”