Trump Banned the Military’s Best AI Because It Refused to Enable Surveillance

On February 27, 2026, President Trump ordered every federal agency to stop using Anthropic after the company refused one Pentagon demand: drop the restrictions preventing its AI from being used for mass domestic surveillance and autonomous weapons that fire without human approval. The Pentagon threatened to invoke the Korean War-era Defense Production Act to forcibly strip those limits away. It also threatened to label Anthropic a “supply chain risk,” a designation normally reserved for adversaries like China’s Huawei. Trump called the company “leftwing nut jobs” on Truth Social and issued the ban the same afternoon. OpenAI’s CEO publicly said his company holds the same position on surveillance and autonomous weapons. Over 100 Google employees signed a petition asking their company to follow suit. Claude is currently the only AI cleared for classified military systems, including the operation that captured Nicolás Maduro. Pentagon officials privately admitted replacing it would be “a huge pain in the ass.” The administration just degraded real military capability to punish a company for drawing a line against domestic spying. The government should use whatever tool works best for the mission, regardless of whether the president likes the company behind it. Senators Ed Markey and Chris Van Hollen called these threats “a chilling abuse of government power.” I’m asking you to co-sponsor legislation restricting executive use of the Defense Production Act against domestic companies outside genuine wartime emergencies. Then demand a public congressional hearing within 30 days on what the Pentagon actually means by “all lawful purposes.” Congress wrote these laws. Start enforcing them.