“Claude Code became the reference choice for programmers from all over the world and therefore it was not a surprise to learn the technical staff of Openi also used our coding tools before the launch of GPT-5,” said the anthropic spokesman Christopher Nulty in a declaration at Wired. “Unfortunately, this is a direct violation of our terms of service.”
According to the anthropic Terms of commercial serviceCustomers are prohibited from using the service to “build a competing product or service, also to form artificial intelligence models” or “Reverse engineer or duplicate” services. This change in the access of Openai to Claude comes as the chatgpt is According to reports prepare to release a new AI model, GPT-5, which is he said To be better in coding.
Openi was connecting Claude to his internal tools using Special Developer Access (API), instead of using the normal chat interface, according to sources. This allowed the company to carry out tests to evaluate Claude’s skills in things such as coding and creative writing against its artificial intelligence models and verify how Claude has responded to security instructions that involve categories such as CSAM, self -harm and defamation, the sources say. The results help to compare the behavior of their models in similar conditions and make necessary changes.
“It is standard of the sector to evaluate other artificial intelligence systems to compare progress and improve safety. While we respect the anthropic decision to cut our API access, it is disappointing considering that our API remains available to them,” said the director of the Chief Communications of Openi, Hannah Wong in a statement to Wired.
Nulty says that Anthropic “will continue to ensure that Openi has API access for the purposes of benchmarking and security assessments such as standard practice throughout the sector”. The company did not respond to the request for clarifications of Wired on itself and how the current restriction of Claude API of Openai would have had an impact on this work.
The best technological companies that illustrate the API access from competitors was a tactic in the technological sector for years. Facebook he did the same with the screw owned by Twitter (which led to accusations of anti -concurrence behavior) and Salesforce last month limited competitors from access to certain data through the Slack API. This is not even the first for anthropic. Last month, the company limited the direct start of Windsurfing Windsurfing Windsurf to its models after it was said that Openi was destined to acquire it. (That agreement failed).
Jared Kaplan, Chief Science Officer of Anthropic I spoke with techcrunch At the time, revoke Windsurf access to Claude, saying “I think it would be strange for us to sell Claude to Openai”.
A day before cutting Openi’s access to the API Claude, anthropic announced New rate limits on the Claude code, its coding tool based on artificial intelligence, citing Explosive use and, in some cases, Violations of its terms of service.