The US government is imposing strict new rules for civilian AI contracts, requiring companies to grant the US an irrevocable license to use their models for any lawful purpose. This move follows the Pentagon's decision to terminate a $200 million deal with Anthropic after the AI firm refused to allow unfettered government access to its technology. The new guidelines from the General Services Administration (GSA) mandate that contractors provide a neutral, non-partisan tool, explicitly banning the encoding of partisan or ideological judgments. Anthropic, valued at $380 billion, had argued its powerful models could enable domestic mass surveillance if used for all lawful purposes and sought specific safeguards.

The Pentagon accused the company of seeking veto power over military operations. The GSA is soliciting industry feedback before finalizing these rules, which mirror Pentagon considerations for military contracts. Anthropic also faces a supply-chain risk designation, typically reserved for Chinese or Russian firms.