According to a report by Wired, Openai has revoked access to the Claude family of AI models.
Sources told Wired that Openai has connected Claude to internal tools, allowing the company to compare Claude's performance with its own models in categories such as coding, writing and safety.
TechCrunch contacted humanity for comments. In a statement to Wired, a humanity spokesman said, “Openai's own technical staff had used coding tools prior to launching the GPT-5.” (Anthropic's commercial terminology prohibits businesses from using Claude to build competing services.)
However, humanity also said it will continue to provide Openai access for “benchmarks and safety assessments.”
Meanwhile, a spokesperson for Openai, a unique statement, described its usage as “industry standard,” adding, “it's disappointing to consider that APIs will continue to be available, while respecting humanity's decision to block API access.”
Human leaders had already shown resistance to providing access to competitors. Chief Science Director Jared Kaplan previously justified Windsurf (a rumor decision that was a rumored open acquisition target and subsequently acquired by recognition).