Terms of Use and Competitive Boundaries
Anthropic alleges that OpenAI used its Claude models via developer API access to support internal benchmarking and safety evaluations. The terms of service for Claude strictly prohibit the use of its tools to build or train competing models, reverse engineer the service, or develop rival AI systems.
“Claude Code has become the go-to choice for coders everywhere, and so it was no surprise to learn OpenAI’s own technical staff were also using our coding tools ahead of the launch of GPT-5,” said Christopher Nulty, Anthropic’s spokesperson. “Unfortunately, this is a direct violation of our terms of service.”
Although OpenAI was reportedly not using Claude via the public chat interface, it was integrating the models into its internal systems through special API access. This allowed the company to analyze Claude’s performance in areas such as code generation, creative writing, and safety prompts related to harmful content. The goal was to compare Claude’s behavior with OpenAI’s own models under identical conditions.
OpenAI Responds
While acknowledging the cutoff, OpenAI emphasized that such cross-evaluations are a standard practice in the AI industry.
“It’s industry standard to evaluate other AI systems to benchmark progress and improve safety,” said Hannah Wong, OpenAI’s chief communications officer. “While we respect Anthropic’s decision to cut off our API access, it’s disappointing considering our API remains available to them.”
Despite the restriction, Anthropic indicated it still intends to allow limited access to OpenAI for the purpose of benchmarking and safety checks. However, the company did not clarify how this would be implemented without API access.
Competitive Tensions Escalate
This isn’t the first time Anthropic has restricted access to its models amid competitive concerns. In July, it revoked access to the AI coding startup Windsurf after rumors emerged that OpenAI was in talks to acquire the company—a deal that ultimately fell through. At the time, Anthropic’s chief science officer Jared Kaplan told TechCrunch, “I think it would be odd for us to be selling Claude to OpenAI.”
The decision to restrict OpenAI comes just one day after Anthropic introduced new rate limits on Claude Code, its AI coding tool, attributing the move to surging demand and occasional breaches of its service terms.
A Broader Industry Pattern
Anthropic’s move reflects a broader trend in the tech industry, where major players increasingly shut out rivals from APIs and data pipelines to maintain a competitive edge. Facebook infamously pulled the plug on Vine’s access to its social graph, and more recently, Salesforce limited competitor access to Slack APIs.
As AI development accelerates and the stakes rise, boundaries around proprietary tools and data are becoming flashpoints in the race to dominate the future of artificial intelligence. The OpenAI-Anthropic API clash may not be the last of its kind.