The Pentagon says it has stopped using Anthropic’s Claude AI, ending a months-long dispute over military access to the powerful AI model.
The fight centered on a major question. How much control should an AI company have over the way the U.S. military uses its technology?
Anthropic allowed the Pentagon to use Claude for many national security purposes. However, the company maintained restrictions on certain applications.
Those restrictions included mass domestic surveillance and fully autonomous weapons.
Pentagon officials wanted broader access to commercially available AI systems. Anthropic refused to remove its remaining safeguards.
Now, the Pentagon says its use of Anthropic products has ended.
How the Pentagon and Anthropic Reached This Point
Anthropic has worked with the U.S. government on national security applications for years.
Claude supported a wide range of defense-related tasks. These included research, intelligence analysis and other military applications.
However, Anthropic drew firm boundaries around some uses of its technology.
The company opposed using Claude for mass surveillance of Americans. It also restricted the use of its technology for fully autonomous lethal weapons.
That position eventually collided with the Pentagon’s approach to artificial intelligence.
Defense officials wanted AI companies to allow the military to use their systems for lawful purposes. Anthropic argued that developers should still maintain certain safety restrictions.
Neither side backed down.
Anthropic Refused to Remove Key AI Safeguards
Anthropic CEO Dario Amodei defended the company’s restrictions.
The company would not grant unrestricted access for autonomous weapons or mass domestic surveillance.
The Pentagon saw the issue differently.
Military officials argued that private technology companies should not determine which lawful national security operations the government can pursue.
That disagreement quickly became much bigger than a contract dispute.
It turned into a debate about who ultimately controls powerful artificial intelligence.
Pentagon Moves Away From Anthropic
The conflict escalated when Defense Secretary Pete Hegseth moved against Anthropic.
The Pentagon designated the company as a national security supply-chain risk. Anthropic challenged that decision in court.
In September, a federal appeals court upheld the Pentagon’s designation in a 2-1 decision.
At the same time, the Pentagon started moving its AI workloads away from Anthropic.
Under Secretary of Defense for Research and Engineering Emil Michael said in September that the military had already transitioned roughly 90% of its classified systems away from Anthropic.
Officials expected to finish that transition by the end of the month.
The Pentagon now says it has ceased using Anthropic products.
Claude Had Become Part of the Pentagon’s AI Infrastructure
Removing Claude was not as simple as deleting a chatbot.
The technology had reportedly become part of the Pentagon’s broader AI infrastructure.
Reports indicate that Claude worked with Palantir’s Maven Smart System. Maven helps military personnel process and analyze large amounts of intelligence data.
Claude reportedly supported research, analysis and intelligence-gathering operations.
That integration helps explain why the Pentagon needed time to move away from Anthropic.
Modern AI models increasingly operate as infrastructure rather than standalone applications.
Organizations connect them to databases, software platforms and internal workflows. Replacing one model can require major changes throughout an entire technology system.
For the Pentagon, switching AI providers can become a major technical operation.
The Bigger Fight: Who Controls Military AI?
The Pentagon-Anthropic dispute could set an important precedent for the AI industry.
Anthropic believes developers should maintain some boundaries around their technology. That position applies even when the customer is the U.S. military.
The Pentagon raises a different question.
Should a private technology company decide how the U.S. government can use legally purchased technology?
That debate will become increasingly important.
Artificial intelligence already supports intelligence analysis, cybersecurity and surveillance. Future systems could play much larger roles in targeting and autonomous military operations.
That means AI companies may increasingly influence decisions once controlled almost entirely by governments.
The Pentagon Is Not Walking Away From AI
The Pentagon’s split with Anthropic does not represent a retreat from artificial intelligence.
Instead, the military continues to accelerate its adoption of AI.
The Defense Department has worked with several major technology companies and AI developers. It also continues to expand AI across national security operations.
The United States sees artificial intelligence as a critical technology in its competition with global rivals.
That makes this dispute especially important.
The Pentagon did not reject AI.
It rejected Anthropic’s terms for using it.
A Warning for the Entire AI Industry
Other AI companies will closely watch what happens next.
Every major developer could eventually face the same dilemma.
What happens when a company’s safety rules conflict with the government’s definition of lawful use?
Can an AI company restrict military applications after its technology becomes deeply embedded in national security systems?
Should governments rely on AI models when private companies can impose their own restrictions?
And who gets the final word when national security and AI safety collide?
Those questions are no longer theoretical.
The Pentagon and Anthropic have turned them into a real-world test.
For now, the Pentagon says Claude is out.
But the much bigger battle over who controls military AI is only beginning.
