Anthropic CEO Responds to Pentagon Ban, Opposes AI Surveillance and Fully Autonomous Weapons

Anthropic has supported the U.S. military since June 2024 and secured a Pentagon contract worth up to $200 million in July 2025, with Claude also deployed on classified networks. At the heart of the dispute is the Pentagon’s demand that the model be available for all lawful uses. Anthropic insists that it must not be used for mass surveillance of Americans or fully autonomous weapons that can decide to fire without human involvement.

On February 24, 2026, Defense Secretary Pete Hegseth ordered CEO Dario Amodei to withdraw the restrictions by February 27 or face contract termination, designation as a supply-chain risk or the use of the Defense Production Act. The Pentagon formally notified Anthropic of the designation on March 4. Anthropic filed two lawsuits on March 9, saying several hundred million dollars in 2026 defense revenue was at risk and the broader impact could run into billions of dollars.

On July 14, 2025, the U.S. Department of Defense awarded Anthropic a contract worth up to $200 million to deploy Claude in classified national-security and military operations. The dispute is not over whether Anthropic should support national defense, but over the military’s demand that Claude be available for “all lawful uses.” Anthropic argues that existing laws may not keep pace with AI risks and insists on excluding domestic mass surveillance and fully autonomous weapons that operate without human involvement.

Defense Secretary Pete Hegseth issued an ultimatum on February 24, 2026, demanding that Anthropic remove the restrictions by 5:01 p.m. on February 27. After negotiations collapsed, the Pentagon formally designated Anthropic a supply-chain risk on March 5, putting the $200 million contract at risk of termination. Chief Executive Dario Amodei has reopened talks but maintains that Claude is not yet reliable enough for military defense, and is seeking explicit terms that preserve the company’s two red lines.

Anthropic’s Claude has been incorporated into US Department of Defense workflows for intelligence analysis and target identification. The company secured a contract worth up to $200 million in July 2025. The Trump administration later ordered federal agencies to stop using Claude, citing national security risks and failed contract negotiations, highlighting tensions between commercial AI and military requirements.

The US military continued using Claude to analyze intelligence and identify targets during airstrikes on Iran just hours after Trump issued the ban, The Wall Street Journal reported. The operation struck about 1,000 targets in its first 24 hours. The report did not specify the exact date of the order or disclose how much money was involved in the mission.