The Trump administration has added Anthropic to a federal blacklist, revoking its eligibility for Pentagon contracts after the AI firm refused to comply with national security data-sharing demands. The move marks a significant escalation in government-industry tensions over AI governance.
- Anthropic excluded from all DoD contracts effective February 2026
- Pentagon required real-time access to model weights, training data, and logs
- Anthropic’s $120M in 2025 federal contract revenue suspended
- DoD cites national security concerns as primary justification
- Other firms like Palantir and IBM remain compliant and active
- No public disclosure of additional firms under review
The Department of Defense has formally excluded Anthropic from participating in classified defense technology initiatives, citing the company’s refusal to meet requirements for model transparency and data access under national security protocols. The decision, effective immediately, bars the San Francisco-based firm from all future Defense Advanced Research Projects Agency (DARPA) and Joint All-Domain Command and Control (JADC2) programs. The exclusion follows a directive issued in late February 2026 requiring AI developers receiving federal funding or working on defense systems to grant the U.S. government real-time access to model weights, training data, and deployment logs. Anthropic rejected the terms, arguing they would compromise user privacy and intellectual property, and instead proposed a limited data audit framework. The administration deemed this insufficient. As a result, Anthropic’s revenue from federal contracts—estimated at $120 million in 2025—has been suspended. The company had been a key participant in the Defense Innovation Unit’s AI pilot programs for surveillance and logistics optimization. Other firms, including Palantir and IBM, continue to hold active defense contracts under the same conditions. The move has triggered backlash within Silicon Valley, where several tech leaders have criticized the government’s approach as overly prescriptive. Meanwhile, defense contractors reliant on AI for predictive maintenance and threat detection now face potential delays in project timelines. The Pentagon has not disclosed whether other AI firms are under similar review.