Pentagon Testing AI Alternatives to Anthropic Claude
- •The Pentagon is testing new AI models from OpenAI and Google to replace Anthropic's Claude.
- •The shift follows a March 2026 supply chain risk designation by Defense Secretary Pete Hegseth.
- •Pentagon officials are working with 25 internal power users to evaluate alternative AI system replacements.
The US Department of Defense has initiated testing of new AI models from providers including OpenAI and Google as it seeks to replace Anthropic's Claude. This transition follows a public fallout between the defense agency and the firm led by Dario Amodei regarding the use of unrestricted AI capabilities. As a result of this dispute, Anthropic withdrew from a multi-million dollar contract previously established to integrate its models into Pentagon operations.
The shift began in March 2026, specifically three days after US Defense Secretary Pete Hegseth officially designated Anthropic as a "supply chain risk." To facilitate this transition, the Pentagon is actively working with 25 designated "power users" within the organization to evaluate alternatives to the Claude platform. While reports indicate that a new agreement with OpenAI has been signed, the department is currently conducting a broader search to identify suitable replacements for its existing AI-integrated defense systems.