Anthropic has cut off OpenAI’s API access to its Claude models after determining that OpenAI violated Anthropic’s terms of service. According to Anthropic, OpenAI used Anthropic’s developer APIs internally to test and compare Claude’s capabilities in tasks like coding and safety. These tests were part of OpenAI’s own work on new AI models like GPT-5.
Anthropic’s commercial terms dictate that customers cannot use its services to create or train competing AI models, or to reverse engineer or duplicate them. The company crossed a line as OpenAI integrated Claude into its own systems to evaluate and benchmark the model, especially in categories like code generation, creative writing, and how Claude responded to safety-related prompts, Anthropic claims. Anthropic says that OpenAI will still be able to use the Claude API for benchmarking and safety testing, as is standard practice.
Anthropic spokesperson Christopher Nulty explained that the brand’s coding tool, Claude Code, has become a go-to option for many developers and confirmed that OpenAI’s activity directly violated its policies.
OpenAI’s communications head, Hannah Wong, responded that it is common across the industry to evaluate other companies’ AI tools to compare progress and improve safety. She added that OpenAI’s APIs remain available to Anthropic for similar testing.
Anthropic has experience limiting API access to competitors, as seen last month when the company cut off the AI coding startup Windsurf over similar concerns, as reported by Wired.
© 2001-2025 Ziff Davis, LLC., a Ziff Davis company. All Rights Reserved.
ExtremeTech is a federally registered trademark of Ziff Davis, LLC and may not be used by third parties without explicit permission. The display of third-party trademarks and trade names on this site does not necessarily indicate any affiliation or the endorsement of ExtremeTech. If you click an affiliate link and buy a product or service, we may be paid a fee by that merchant.
