According to a recent report by The Wire, Anthropic has terminated OpenAI's access to its Claude series AI models.

Industry sources revealed to the publication that OpenAI was integrating Claude with internal tools to benchmark the model's performance against its own offerings in coding, writing, and security categories.

In a statement to TechCrunch, an Anthropic spokesperson stated: "OpenAI engineers had been using our coding tools prior to GPT-5's launch, which clearly violates our service terms." The company's commercial policy explicitly prohibits organizations from using Claude to develop competing services.

However, Anthropic confirmed it will continue granting OpenAI limited access for "benchmarking and security evaluation purposes."

OpenAI's representative described their usage as "industry standard" and added: "While we respect Anthropic's decision to cut our API access, it's disappointing considering our APIs remain available for their use."

Anthropic executives have consistently resisted providing access to competitors. Chief Science Officer Jared Kaplan previously justified ending connections with Windsurf (the rumored OpenAI acquisition target later acquired by Cognition), stating: "I find it strange to sell Claude to OpenAI."