Anthropic wins US court fight over Pentagon supply chain blacklist

Anthropic just won a major constitutional and commercial reprieve. Being labeled a supply chain risk and effectively cut off from US government and defense contracts would have pushed one of the leading alignment-focused labs to the margins of the state’s AI ecosystem. The court’s finding that the Pentagon’s measures were “illegal and baseless” resets that relationship and signals to other labs that push back on military use that they are not automatically risking blacklisting.

Strategically, this keeps Anthropic in the center of the US AI-industrial complex rather than forcing it into an oppositional niche. That matters because access to defense, intelligence and broader federal workloads is increasingly where serious long-horizon AI safety, evals and red teaming get funded. It also sets a precedent that public criticism of national security AI policy is protected speech, not grounds for exclusion from procurement.

For the race to AGI, the ruling does not change model scaling limits overnight, but it does preserve competitive balance among frontier labs. A sidelined Anthropic would have left more room for OpenAI, Google and others to dominate defense-linked contracts and steer safety norms. With Anthropic back at the table, expect a harder-edged debate over how much autonomy and lethal capability governments should be allowed to build on top of frontier systems.