Anthropic blocks OpenAI’s access to Claude AI models, here's why

Anthropic

Anthropic has reportedly cut off OpenAI’s access to its Claude family of AI models, according to Wired. The move follows allegations that OpenAI used Claude with internal tools to compare model performance in areas such as coding, writing, and safety.

A TechCrunch report quotes an Anthropic spokesperson stating that OpenAI’s technical team had also used Claude’s coding capabilities in the lead-up to GPT-5’s release — a move described as “a direct violation of our terms of service.” Anthropic’s commercial policy clearly prohibits using Claude to build rival services.

Despite the breach, Anthropic says it will still allow limited access for “benchmarking and safety evaluations.”

Responding to the decision, an OpenAI spokesperson called the usage “industry standard” and expressed disappointment: “While we respect Anthropic’s decision to cut off our API access, it’s disappointing considering our API remains available to them.”

Anthropic has previously been cautious about granting access to potential competitors. Chief Science Officer Jared Kaplan once defended the decision to block Windsurf — a startup later acquired by Cognition — by saying, “I think it would be odd for us to be selling Claude to OpenAI.”