Anthropic revoke OpenAI’s access to its Claude AI models

Just a heads up, if you buy something through our links, we may get a small share of the sale. It’s one of the ways we keep the lights on here. Click here for more.

Anthropic has officially stopped OpenAI from using its Claude models. 

This decision came after reports revealed that OpenAI had been using Claude in ways that Anthropic says broke the rules.

According to a report by Wired, OpenAI had connected Claude to some of its internal tools. 

These tools were used to compare how well Claude performed against OpenAI’s own AI models, especially in areas like computer programming, writing, and safety. 

Anthropic wasn’t happy about this because their commercial agreement clearly says that companies are not allowed to use Claude to help develop competing AI services.

Anthropic released a statement to clarify its stance. 

They said that OpenAI’s technical team had also been using Claude’s coding features right before the launch of OpenAI’s next big AI model, GPT-5. Anthropic views this as a clear violation of its terms of service. 

These terms are meant to stop companies like OpenAI from using Claude in a way that could benefit their own competing products.

However, Anthropic isn’t shutting OpenAI out completely. 

They said they will still allow OpenAI to access Claude’s models, but only for specific tasks, like benchmarking (which means testing and comparing models) and safety evaluations.

OpenAI responded by saying that its use of Claude was completely standard in the industry. 

They expressed disappointment over the decision, especially since OpenAI’s own AI services remain open to Anthropic if it wants to use them.

This isn’t the first time Anthropic has denied access to a competitor. One of its top executives, Jared Kaplan, had previously said it didn’t make sense to let OpenAI use Claude. 

He made this statement in the context of OpenAI being linked to Windsurf, a company that was later bought by another AI startup called Cognition.

This situation highlights rising tensions in the competitive world of AI, where companies are becoming more protective of their tools and data.

Do you think AI companies should restrict competitors from using their models? Or does this create an unhealthy closed ecosystem that could hurt innovation and development? Tell us below in the comments, or reach us via our Twitter or Facebook.