Google said it would continue to offer Anthropic's artificial intelligence technologies to its customers for uses not linked to defence. The decision follows the US Department of Defense classifying the start-up as a supply-chain risk and after Microsoft took a similar stance.

Anthropic's Claude models will remain available on Google Cloud via the Vertex AI platform. Google is one of the start-up's main investors, with total commitments of $3bn after an additional $1bn investment announced in January 2025. Anthropic also uses Google Cloud infrastructure to train its models and has access to up to 1 million tensor processing units (TPUs).

The controversy erupted after Anthropic refused to accept certain terms of use imposed by the Pentagon. The Trump administration then ordered federal agencies to stop using its technologies and began a phased withdrawal over six months. Chief executive Dario Amodei said the company will challenge the decision in court, while several defence-sector companies are now turning to alternatives such as OpenAI's.