WASHINGTON (AP) — The U.S. Defense Department and Anthropic, a leading artificial intelligence company, are locked in a high-stakes standoff over the military’s use of the AI model Claude, with the Pentagon issuing an ultimatum that could reshape the future of AI in defense. The dispute, escalating rapidly this week, centers on Anthropic’s concerns about the potential for its technology to be used in the development of autonomous weapons systems or for mass surveillance.
Defense Secretary Pete Hegseth demanded Friday that Anthropic grant the military unfettered access to Claude, according to a senior Pentagon official. The official stated Anthropic has until 5:01 p.m. to comply, or the Pentagon will invoke the Defense Production Act – a 1950 law allowing the government to compel private sector cooperation for defense purposes – or designate Anthropic as a supply chain risk, effectively barring other defense contractors from using its products.
Anthropic, which signed a $200 million deal with the Pentagon last July, is currently the only AI company whose technology is used in classified military systems. The company wants guarantees its models won’t be used for autonomous weapons or surveillance, a stance that reportedly flared up after reports surfaced that Claude was used in the January raid to capture Venezuelan leader Nicolás Maduro through a partnership with Palantir. Anthropic CEO Dario Amodei reportedly told Hegseth the company has never interfered with legitimate military operations, but reiterated the two red lines regarding autonomous weapons and surveillance.
The Pentagon insists it should be allowed to use Claude within legal boundaries. Pentagon spokesperson Sean Parnell posted on X that the Defense Department “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. This narrative is fake and being peddled by leftists in the media.” The Pentagon’s directive on autonomous weapons, updated in 2023, calls for “appropriate levels of human judgment over the use of force.”
The standoff highlights a broader debate over the ethical boundaries of AI in warfare and the extent to which technology companies can control how their products are used by the military. Luke Barnes, a senior research scientist at New York University’s Stern Center for Business and Human Rights, warned that prioritizing “immediate tactical advantages” could lead to a “race to the bottom” in ethical standards, potentially impacting global dynamics.
Elon Musk’s xAI has already agreed to allow its Grok model to process classified information under Pentagon conditions, and the military also uses Google’s Gemini and OpenAI’s ChatGPT in unclassified settings.
Meanwhile, U.S.-Iran nuclear talks resumed Thursday in Geneva, with Oman’s foreign minister reporting “significant progress.” The United States has reportedly presented Iran with stringent demands, including the destruction of its three main nuclear sites and the handover of enriched uranium. The U.S. has also increased its military presence in the Middle East, raising concerns about potential military action.
Democrats in the House and Senate are pushing legislation to require congressional approval before any military action against Iran, but have previously failed to limit the president’s war powers.
Adding to the geopolitical tension, Borge Brende resigned Thursday as president of the World Economic Forum amid scrutiny over his ties to convicted sex offender Jeffrey Epstein.
On Monday, March 2, Melania Trump is scheduled to chair a session of the United Nations Security Council as the United States assumes its rotating presidency, a first for a U.S. first lady. Secretary of State Marco Rubio is also expected to meet with Israeli Prime Minister Benjamin Netanyahu in Israel. Nepal is preparing to hold national elections on March 5.
