Pentagon Pressures Anthropic on AI Safety
The U.S. Department of Defense is reportedly pressuring AI firm Anthropic to relax safety guardrails in its Claude models for military applications. The standoff raises fundamental questions about AI autonomy in warfare and the ethics of dual-use technology. The outcome could set a precedent for how AI developers balance commercial, regulatory, and defense sector demands.
The Pentagon's demand centers on "any lawful use" of Anthropic's AI, a broad term the company fears could override its specific prohibitions against use in fully autonomous weapons and for mass domestic surveillance of American citizens. CEO Dario Amodei stated the company "cannot in good conscience accede" to the request, arguing that current AI is not reliable enough for autonomous targeting and that AI-driven surveillance poses "serious, novel risks to our fundamental liberties." In response to Anthropic's refusal, the Department of Defense has threatened to terminate its contracts, designate the company a "supply chain risk," and potentially invoke the Defense Production Act, a Cold War-era law that would give the government more direct control over the company's technology. The Pentagon has given Anthropic a deadline of 5:01 p.m. ET on Friday, February 27, to comply. This public conflict follows a January 2026 memorandum from the Department of Defense outlining a new, accelerated "AI-first" strategy aimed at rapid experimentation and deployment of AI capabilities. Anthropic was awarded a contract with a ceiling of $200 million in July 2025 and, until recently, was the only major AI model deployed across a number of the military's classified systems. The standoff highlights a growing rift within the tech sector. While Anthropic holds its position, competitors like Google, OpenAI, and xAI are also negotiating with the Pentagon. Reports suggest these other firms have been more accommodating of the military's terms. However, the dispute has sparked internal dissent, with hundreds of employees at Google and OpenAI signing open letters in solidarity with Anthropic, urging their own leadership to establish similar "red lines" in military contracts. Adding another layer to the situation, Anthropic recently overhauled its own "Responsible Scaling Policy." The company removed a key pledge to halt the development of new models if it couldn't guarantee safety mitigations in advance, citing the rapid pace of competitors and a lack of government regulation. While Anthropic states this change is unrelated to the Pentagon dispute, it signals a shift in the company's approach to self-regulation amid intense market and geopolitical pressures. Internationally, the debate over "killer robots" continues, with the United Nations General Assembly passing a resolution in December 2024 calling for new regulations on lethal autonomous weapons systems (LAWS). However, no consensus for a binding international treaty has been reached, leaving the development and deployment of such technologies in a contentious legal and ethical gray area.
