Anthropic Faces Pentagon Showdown as Pete Hegseth Threatens Contract Termination Over AI Guardrails
WASHINGTON — A high stakes standoff is unfolding between one of America’s leading artificial intelligence firms and the Pentagon, raising urgent questions about who controls the ethical boundaries of military AI.
Defense Secretary Pete Hegseth has given Anthropic CEO Dario Amodei a Friday deadline to loosen safeguards on the company’s AI model, Claude or risk losing a $200 million Defense Department contract. According to multiple sources familiar with the discussions, Hegseth also threatened to designate Anthropic a “supply chain risk” and invoke the Defense Production Act (DPA) to compel cooperation.
At the center of the dispute: whether Anthropic will remove restrictions preventing its AI from being used in autonomous weapons systems or for mass domestic surveillance.
The Pentagon’s Demand: “All Lawful Use”
The Pentagon wants Anthropic to lift internal restrictions so Claude can be deployed for “all lawful use,” according to two sources familiar with the negotiations.
A Pentagon official told CNN that the company has until 5:01 p.m. Friday to “get on board or not.” If Anthropic refuses, the official said Hegseth would move to terminate the contract and could invoke the DPA, a Cold War-era law that allows the federal government to direct private companies in the interest of national defense. The official added that Anthropic could also be labeled a supply chain risk, a designation that would effectively bar companies with military contracts from using Anthropic’s products in defense-related work. Such a designation is typically reserved for companies viewed as extensions of foreign adversaries such as Russia or China.
Anthropic’s Red Lines
Anthropic has made clear it will not remove two specific guardrails:
• Use of AI in fully autonomous weapons systems
• Use of AI in mass domestic surveillance of U.S. citizens
Sources familiar with the company’s position say Anthropic believes current AI systems are not reliable enough to control lethal weapons without meaningful human oversight. The company also argues there is no comprehensive legal framework governing large scale AI-driven surveillance. Anthropic has publicly positioned itself as the most safety-focused of major AI firms. Its founders, former employees of OpenAI, left over disagreements regarding AI development speed and safety controls. The company recently committed $20 million to a political initiative advocating stronger AI regulation.
A Legal and Strategic Contradiction?
Katie Sweeten, a former Justice Department liaison to the Defense Department, questioned how the Pentagon could simultaneously label Anthropic a supply chain risk and compel it to work with the military under the DPA.
“I would assume we don’t want to utilize the technology that is the supply chain risk, right? So I don’t know how you square that,” Sweeten told CNN. “What it sounds like is that the supply chain risk may not be a legitimate claim, but more punitive because they’re not acquiescing.”
The Pentagon disputes the characterization that this dispute centers on autonomous weapons or surveillance. A defense official told CNN the matter “has nothing to do with mass surveillance and autonomous weapons being used” and emphasized that legality is the Pentagon’s responsibility as the end user.
“You can’t lead tactical ops by exception,” the official said. “Legality is the Pentagon’s responsibility.”
A Cordial Meeting, High Stakes Outcome
Despite the public tension, sources say Tuesday’s meeting between Hegseth and Amodei was cordial. No raised voices. No visible animosity. Anthropic described the meeting as a “good-faith” discussion.
“Anthropic CEO Dario Amodei met with Secretary Hegseth at the Pentagon this morning,” the company said in a statement. “We continued good-faith conversations about our usage policy to ensure Anthropic can continue to support the government’s national security mission in line with what our models can reliably and responsibly do.”
Negotiations have reportedly been ongoing for months, with tensions escalating after reports surfaced last week that Hegseth was preparing to cut the contract.
The Competitive Implications
Anthropic’s resistance could open doors for competitors. The Pentagon confirmed that xAI, founded by Elon Musk, is “on board with being in a classified setting,” while other AI firms are reportedly close to similar arrangements.
For Anthropic, being labeled a supply chain risk could severely damage its enterprise ambitions. Many major corporations maintain defense contracts; a formal designation could deter them from adopting Anthropic’s products. At stake is not only $200 million in Pentagon funding but broader credibility in government and enterprise markets.
The Bigger Question: Who Sets the Limits?
The confrontation underscores a deeper structural issue in the AI era: who determines the ethical boundaries of machine intelligence when national security is involved? If the Pentagon invokes the Defense Production Act to override Anthropic’s guardrails, it would mark one of the most aggressive federal interventions into private AI governance to date. If Anthropic holds its ground, it could redefine the relationship between Silicon Valley and the defense establishment, signaling that AI firms may assert independent red lines even under federal pressure.
The deadline looms. The outcome will not just shape one contract. It could shape the future architecture of military AI in the United States.
