Anthropic has revoked OpenAI's access to its Claude family of artificial intelligence models.
As Wired reports, the reason was OpenAI's use of Claude tools for internal performance comparisons with its own models, particularly in the areas of programming, writing, and security.
An Anthropic representative told TechCrunch that OpenAI’s technical staff used these tools before launching GPT-5, which is against the company’s terms of service. Anthropic’s commercial terms prohibit using Claude to build competing services.
Despite this, Anthropic has left the ability for OpenAI to evaluate Claude's models for benchmarking and security analysis purposes. In response, an OpenAI spokesperson called this use "industry standard" and expressed disappointment with Anthropic's decision, noting that OpenAI's API remains available to Anthropic.
By the way, it was previously reported that Anthropic's artificial intelligence models turned out to be the most popular among business users, while ordinary users prefer OpenAI models.