Sen. Elizabeth Warren, D-Mass., is asking Defense Secretary Pete Hegseth to explain why the Pentagon branded Anthropic as a “supply chain risk,” claiming that the move appears to be retaliatory.
On March 23, Warren sent a letter to Hegseth seeking insight into the Department of Defense’s (DOD) decision to designate Anthropic a “supply chain risk” after the company refused to remove certain guardrails on its artificial intelligence (AI) technology for military use.
According to Warren, the department’s decision appears to be retaliatory. She requested that Hegseth “immediately provide further details regarding whether DOD is weaponizing bipartisan national security statutes to retaliate against an American company.”
Under the Trump administration, the DOD has been rebranded as the Department of War.
DOD v. Anthropic: Where it all began
If you have not been following along with this saga, here is a recap.
The department’s move to cut ties with Anthropic came amid a broader dispute with the company over its stance against surveillance and autonomous weapons. Pentagon officials argued that the department wants to be able to use the AI without “having to call Anthropic for permission to shoot down enemy drone swarms that target Americans,” not conduct mass surveillance.
The conflict escalated after President Donald Trump ordered federal agencies to terminate contracts with Anthropic. The Pentagon then labeled the company a “supply-chain risk,” a designation typically reserved for foreign entities deemed to pose national security threats.
Following the designation, DOD components and the military services were given 180 days to remove Anthropic products from their systems, and defense contractors must certify they are no longer using the company’s AI technology.
Anthropic filed two lawsuits alleging illegal retaliation by the federal government. The federal government defended the Pentagon’s decision to blacklist Anthropic, arguing in a court filing that the move was lawful and driven by national security concerns.
Anthropic and the federal government are set to appear before a judge today for a preliminary hearing in the U.S. District Court for the Northern District of California.
Warren’s investigation
Warren raised concerns that the department’s decision “appears to be an unprecedented abuse of longstanding national security authorities that undermines DOD’s credibility in using these authorities for legitimate purposes in the future.”
Warren also points out that during negotiations, Hegseth warned he would invoke the Defense Production Act (DPA) to compel Anthropic to provide its services to the department. The law is intended to enable the government to mobilize the country’s resources during national emergencies.
The two statutes, Warren claims, are “inherently contradictory.”
“To leverage the DPA would suggest that Anthropic’s services are essential to our country’s national security, while designating the company as a supply chain risk suggests that DOD views the technology as a threat to our country’s national security,” she wrote, adding that this “combined with President Trump’s ideological criticism of Anthropic, suggests that the [DOD’s] actions are motivated more by politics rather than national security.”
Warren also said the dispute raises broader concerns about how the department is deploying AI.
“I am particularly concerned that DOD is trying to strong-arm American companies into providing the Department with the tools to spy on American citizens and deploy fully autonomous weapons without adequate safeguards,” she wrote.
Specifically, she pointed to the department’s agreement with OpenAI, signed the same day it moved to designate Anthropic as a supply chain risk.
OpenAI said in its initial announcement that it was confident the department would not use its AI systems for mass surveillance or fully autonomous weapons because of its “safety stack,” existing laws, and contract language. Warren is not convinced.
As part of her investigation, Warren is also seeking answers from OpenAI CEO Sam Altman. In a separate letter, she requested details about the company’s agreement with the department.
Warren explained that according to legal experts, OpenAI’s contractual language does not do enough to constrain the DOD’s use of its technology.
“As long as the Department can make a plausible case that its conduct is not ‘unlawful,’ it appears to be free to use [OpenAI] models in a manner that harms civilian populations and endangers civil liberties,” Warren wrote.
Warren asked both Hegseth and Altman to respond to her questions by April 6.