Anthropic has blocked OpenAI from accessing its Claude models via API, citing a breach of contract just as GPT-5 approaches release.
According to WIRED, Anthropic says OpenAI's technical teams used Claude Code for internal testing ahead of GPT-5. Anthropic's commercial terms of use explicitly prohibit reverse engineering and using the models to develop competing products.
Anthropic spokesperson Christopher Nulty told WIRED, "it was no surprise to learn OpenAI's own technical staff were also using our coding tools ahead of the launch of GPT-5."
WIRED reports that OpenAI integrated Claude into internal tools via developer access and systematically compared Claude with its own AI models. The tests are said to have included everything from coding and creative writing to security-sensitive content, including CSAM, self-harm, and defamation.
Anthropic says it will continue to allow API access "for the purposes of benchmarking and safety evaluations," which it describes as standard industry practice. The company did not clarify whether OpenAI will still qualify for this kind of access going forward.
OpenAI spokesperson Hannah Wong called the move "disappointing," but said the company respects Anthropic's decision. She also pointed out that OpenAI hasn't restricted Anthropic's API access in return.
Anthropic pulled Claude access from Windsurf after OpenAI acquisition talks surfaced
This isn't the first time Anthropic has cut off access following OpenAI's moves. In June, Anthropic pulled most Claude access from the coding tool Windsurf after reports surfaced that OpenAI planned to acquire the company. Anthropic co-founder Jared Kaplan pointed to both limited compute resources and strategic concerns: "I think it would be odd for us to be selling Claude to OpenAI," Kaplan said at an industry conference.
The deal ultimately went elsewhere. Windsurf's executives and key talent joined Google, and the startup merged with code software provider Cognition. Either way, OpenAI no longer has access to Anthropic's API.
Competition in code generation appears to be a driving force. Internal reports suggest GPT-5 is aiming to outperform rivals in programming tasks, an area where Claude Code has held the lead so far. Anthropic's recent moves look designed to keep Claude data out of competing models.
