Court Rules Pentagon Can Blacklist Anthropic Over Claude Feature Restrictions

A court has ruled that the Pentagon can blacklist AI company Anthropic after the company declined to enable certain features of its Claude models, according to a report from Ars Technica.

The ruling reportedly centered on arguments that placing excessive constraints on AI models used in military contexts could create operational risks. Judges in the case said that "overly constrained AI models" could cause military operations to fail.

Context of the Dispute

The case reflects broader tensions between AI developers, who often impose usage restrictions and safety guardrails on their models, and government or military customers seeking fuller access to a model's capabilities for defense-related applications.

Anthropic has previously stated policies around how its Claude models can be used, including restrictions relevant to military and defense contexts, as part of its approach to responsible AI deployment.

The specific features Anthropic declined to enable, and the precise scope of the blacklist the Pentagon is now permitted to impose, were not detailed in the available report.

The ruling could have implications for how other AI developers structure their usage policies when engaging with government and military customers going forward.