Defense Secretary Pete Hegseth has given a major AI company until Friday to comply with demands to remove safeguards from its artificial intelligence model or risk losing a significant Pentagon contract. The government also threatened to designate the company a supply chain risk, a move that could severely impact its business.
At the heart of the dispute are the ethical guardrails the company, Anthropic, has built into its AI model, Claude. The Pentagon, which holds a $200 million contract with Anthropic, wants these restrictions lifted to allow the military to use the model for “all lawful use.” However, Anthropic is reportedly unwilling to compromise on two key issues: the use of AI in autonomous weapons systems and mass domestic surveillance of American citizens.
According to sources, Anthropic believes AI is not yet reliable enough to be trusted with weaponry, and there are currently no laws or regulations governing AI’s role in mass surveillance. A Pentagon official stated that if Anthropic does not agree to the terms by the deadline, the Defense Department will terminate its contract. Furthermore, the official threatened to invoke the Defense Production Act, a law that grants the government broad powers to compel companies to act in the name of national defense, effectively forcing Anthropic to cooperate.
The supply chain risk designation would be a major blow, prohibiting other companies with military contracts from using Anthropic’s technology. This designation is typically reserved for firms seen as extensions of foreign adversaries. A legal expert expressed confusion at the strategy, noting the contradiction in simultaneously compelling a company to work with the military while labeling its technology a security risk, suggesting the move may be punitive.
Despite the high stakes, a meeting between Hegseth and Anthropic CEO Dario Amodei was described as cordial and respectful. Hegseth praised the company’s products, while Amodei reiterated the firm’s commitment to its ethical red lines. Anthropic characterized the discussions as “good-faith conversations” aimed at supporting national security in a responsible manner. The Pentagon official maintained that the issue is about lawful use and that the department has always followed the law.
With the Friday deadline looming, Anthropic has stated it has no plans to acquiesce to the Pentagon’s demands, setting the stage for a significant confrontation between a government agency and a leader in the AI industry over the future of ethical artificial intelligence.
