A major confrontation between the Pentagon and artificial intelligence company Anthropic is approaching a critical Friday evening deadline, with significant implications for how AI technology might be used in military operations.
The conflict, which must be resolved by 5:01 p.m. Friday, centers on disagreements about safety restrictions for AI use in warfare and surveillance. Military officials are demanding broader access to the technology, while Anthropic maintains certain protective measures should remain in place.
The Pentagon is seeking unrestricted lawful use of AI systems and has issued business threats against Anthropic if the company refuses to remove additional safety protocols.
Former acting defense secretary Chris Miller described the situation as significant for the future of battlefield AI. “It’s a shot across the bow about the future of artificial intelligence and its use on the battlefield,” Miller said. “The outcome will be an acid test for those companies that claim to want to use AI humanely.”
The extended disagreement has created divisions among industry executives, defense officials, and congressional members regarding whether AI technology should operate without limitations, particularly given Anthropic’s position that the technology isn’t ready for fully independent weapons systems.
Democratic Senator Elissa Slotkin voiced concerns during a Thursday hearing for defense secretary nominees. “The average person does not think we should allow weapons systems to get into war and kill people without a human being overseeing that in some way,” Slotkin stated. She added: “I certainly don’t think any American, Democrat or Republican, wants mass surveillance on the American people.”
Pentagon spokesperson Sean Parnell responded to criticism on social media Thursday, rejecting what he called false narratives. “The Department of War has no interest in using AI to conduct mass surveillance of Americans (which is illegal) nor do we want to use AI to develop autonomous weapons that operate without human involvement,” Parnell wrote.
The Pentagon has established $200 million contract frameworks with leading AI companies over the past year, including Anthropic, OpenAI, and Google. Defense officials are pressuring these firms to abandon their individual usage restrictions in favor of accepting broad lawful-use terms.
Anthropic has maintained firm boundaries regarding military applications of its Claude AI technology, specifically opposing use in autonomous weapons and domestic surveillance programs. The company was the first among major AI firms to handle classified materials through its partnership with Amazon’s cloud services.
Anthropic’s CEO Dario Amodei, who left OpenAI in 2020 due to concerns about AI oversight, has cautioned that artificial intelligence development is outpacing legal frameworks. In a Thursday blog post, he warned that advanced technology could collect diverse information to surveil ordinary citizens without their knowledge.
“Anthropic understands that the Department of War, not private companies, makes military decisions,” Amodei wrote, but noted that AI in certain situations “can undermine, rather than defend, democratic values.”
Following Amodei’s meeting with Defense Secretary Pete Hegseth this week, the Pentagon offered modified contract terms. However, both sides appear to remain deadlocked.
An Anthropic representative said Thursday that overnight contract revisions from the Defense Department “made virtually no progress” and would permit “safeguards to be disregarded at will.”
Significant business consequences hang in the balance for Anthropic. The Pentagon has warned it will end its partnership with the startup and classify it as a supply-chain risk if the company doesn’t accept the military’s demands for unrestricted AI access.
Such a designation, typically reserved for suppliers from hostile nations, would prevent defense contractors from using Anthropic’s AI systems in Pentagon-related work.
This potential setback comes as Anthropic competes aggressively for business and government contracts, with national security representing a key growth area.
The Pentagon has requested assessments from contractors including Lockheed Martin regarding their dependence on Anthropic technology ahead of the possible risk classification. The defense contractor network included approximately 60,000 companies as of 2021, including major publicly-traded corporations.
Military officials have made an additional threat that some legal experts question. “If they don’t get on board, SecWar will ensure the Defense Production Act is invoked on Anthropic,” a senior Pentagon official stated, “compelling them to be used by the Pentagon regardless of if they want to or not.”
