In a move that highlights growing competition — and tension — among top AI companies, Anthropic has revoked OpenAI’s access to its Claude family of models, citing violations of its terms of service.
According to a report from Wired, OpenAI had been using Claude via API integrations with internal tools to benchmark performance and compare Claude to its own models in areas such as code generation, writing quality, and safety. This usage reportedly escalated ahead of the launch of OpenAI’s next major model, GPT-5.
Violation of Terms?
In a statement to TechCrunch, Anthropic confirmed that OpenAI’s technical teams had been using Claude specifically for coding tool evaluations, which the company described as “a direct violation” of its commercial terms. Those terms prohibit customers from using Claude to develop or enhance competing services.
“OpenAI’s own technical staff were also using our coding tools ahead of the launch of GPT-5,” Anthropic stated.
However, Anthropic added that limited access for benchmarking and safety research would still be permitted — signaling that while competitive usage is restricted, collaborative industry evaluation is still valued.
OpenAI Responds: “Industry Standard”
In response, an OpenAI spokesperson described the usage of Claude as “industry standard” and expressed disappointment over Anthropic’s decision:
“While we respect Anthropic’s decision to cut off our API access, it’s disappointing considering our API remains available to them.”
The exchange underscores a growing competitive chill in the AI space, particularly between former collaborators. Notably, both companies have roots tied to former OpenAI researchers and share backers like Amazon and Google.
Context: Claude Access Has Been Restricted Before
This isn’t the first time Anthropic has limited access to Claude for strategic reasons. The company previously cut off Windsurf, a startup later acquired by Cognition, which was also rumored to be an OpenAI acquisition target.
Anthropic’s Chief Science Officer Jared Kaplan explained the rationale at the time:
“I think it would be odd for us to be selling Claude to OpenAI.”
Takeaway: AI Rivalries Are Heating Up
As the race toward ever-more-powerful foundation models intensifies, so do concerns about data usage, model benchmarking, and competitive integrity. With GPT-5 on the horizon and Anthropic rapidly advancing Claude, access to rival models is becoming a strategic line in the sand.
Expect more companies to reevaluate how — and with whom — they share access, even as collaboration and safety research remain critical pillars of AI development.