US federal agencies get access to Claude across all three branches of government

Anthropic and the US General Services Administration announced an expansion of Claude’s availability to all three branches of the federal government — executive, legislative and judicial — under the GSA’s OneGov procurement programme. The deal offered Claude for Enterprise and the FedRAMP High-certified Claude for Government to participating agencies for a nominal fee of $1 for up to a year, with Anthropic providing technical support to help agencies integrate the model into existing workflows.

The arrangement went further than an earlier federal deal OpenAI had reached, which covered only the executive branch; Anthropic’s covered legislative offices and the judiciary as well, extending Claude to a portion of government that had not previously had a comparable AI-vendor arrangement. FedRAMP High certification meant the models were approved for use with sensitive unclassified federal data, a threshold agencies typically require before deploying software outside routine administrative use.

The deal was part of a wider pattern of frontier AI labs pricing government access far below commercial rates to establish themselves inside federal agencies early, a strategy also visible in OpenAI’s and Google’s parallel government offers around the same period. For Anthropic, it extended a relationship with the US government that had already included work with defence and intelligence customers, and gave the company a foothold across the institutions — including Congress and the courts — that would later shape AI-specific legislation and litigation.