Claude Gov Models For U S National Security Customers

Claude Gov adapts Anthropic’s models for work inside classified U.S. national-security environments. In its June 2025 announcement, the company said these restricted-access models were already deployed and had been shaped by government customer feedback. One concrete change is fewer refusals when processing classified information, alongside claimed improvements in understanding intelligence and defense documents. That changes which sensitive workflows the models can assist with. Anthropic says the models underwent its usual safety testing, but the announcement supplies no quantitative evaluation of their operational performance.