Anthropic has revoked OpenAI’s access to its Claude family of AI models, according to a report. Sources revealed that OpenAI was connecting Claude to internal tools to compare its performance against its own models in areas like coding, writing, and safety.
Anthropic confirmed the move, stating that OpenAI’s technical staff had been using its coding tools ahead of the GPT-5 launch, which violated its terms of service. The company’s commercial terms prohibit using Claude to develop competing services. However, Anthropic clarified that it would still allow OpenAI access for benchmarking and safety evaluations.
In response, an OpenAI spokesperson expressed disappointment over the decision, noting that its API remains available to Anthropic.
Anthropic has previously restricted access to competitors. Chief Science Officer Jared Kaplan defended the company’s earlier decision to cut off Windsurf, a rumored OpenAI acquisition target, by stating it would be unusual to sell Claude to OpenAI. Windsurf was later acquired by Cognition.