The Ministry of National Defense effectively issued a final ultimatum to Anthropic over the scope of military use of artificial intelligence (AI) models. It warned that if the company does not agree to the ministry's demands, it could take coercive measures such as designating the company a "supply chain risk" vendor or invoking the Defense Production Act (DPA).
According to the Wall Street Journal (WSJ), Pete Hegseth, U.S. Minister of National Defense, met Anthropic Chief Executive Officer (CEO) Dario Amodei at the ministry on the 24th and demanded that the company agree by 5:01 p.m. on the 27th (local time) to allow the U.S. military to use AI models for "all lawful purposes."
Minister Hegseth was said to have indicated that if Anthropic does not take a more flexible stance, the company could be designated a "supply chain risk" vendor or the DPA could be applied to force cooperation with the Pentagon.
If designated a supply chain risk vendor, all contracts and suppliers transacting with the Ministry of National Defense would be barred from using Anthropic's AI model "Claude" in collaboration with the U.S. military. Because this measure is typically applied to foreign companies linked to adversarial nations such as China, its application to a domestic company would be unusual. The DPA, likewise, is a law that has been used only in essential fields such as energy and health during national emergencies, and applying it to tech corporations is considered rare.
The two sides have clashed over the scope of military use for AI. While the Ministry of National Defense holds that model use should be allowed across all lawful domains, Anthropic is sticking to a "red line" that its models cannot be used for mass surveillance of its own citizens or for fully autonomous weapons.
The latest meeting reportedly did not narrow the differences. CEO Amodei argued that the company's restriction principles do not materially hinder Pentagon operations, but Minister Hegseth was said to have demanded consent for comprehensive use.
Some observers say the dispute could affect the broader AI industry. Anthropic had been the only AI model developer approved for use in U.S. military classified systems, but the Ministry of National Defense is expanding its partnerships to other corporations.
In fact, the Ministry of National Defense recently signed a contract to use xAI, led by Elon Musk, and its model "Grok" in classified systems. xAI reportedly agreed to all lawful military uses.
Some suggest the Donald Trump administration is taking issue with Anthropic's political leanings. However, experts note that if the Ministry of National Defense severs cooperation with Anthropic, it could burden national security and efforts to secure technological competitiveness.
Dean Ball, who served as an AI adviser in the Trump administration, criticized in the WSJ that simultaneously raising both a supply chain risk designation and DPA application is a policy-incoherent response. The fact that "Grok," cited as an alternative, has been embroiled in controversies over inaccuracy and racist content is also cited as a weakness.
A Ministry of National Defense official told the U.S. political news outlet Axios, "Completely excluding and replacing Claude would be a very complicated process."