OpenAI has been cut off from accessing Anthropic’s Claude API after the latter accused OpenAI of breaching its terms of service. The decision comes amid reports that OpenAI was utilizing Claude’s capabilities, particularly in coding and creative writing, as it prepares to release its next AI model, GPT-5.
An Anthropic spokesperson stated that OpenAI’s internal team leveraged Claude’s tools inappropriately prior to the launch of GPT-5, which was identified as a violation of their commercial terms. The service agreement explicitly prohibits customers from using Anthropic’s technology to create competing products or to reverse-engineer its services.
This restriction on OpenAI follows a pattern in the tech industry where firms regularly restrict competitors’ API access. Similar actions have been seen in past instances, such as when Facebook curtailed access to Vine amid accusations of anti-competitive practices. Anthropic recently took similar actions against the AI startup Windsurf for apparent implications of an OpenAI acquisition.
OpenAI had been engaged in testing Claude’s functionalities by integrating it into its internal tools. This approach allowed the company to benchmark Claude’s performance against its own AI systems and adjust its technology based on comparative results, especially in areas concerning safety and ethical implications.
While Anthropic’s representative acknowledged the decision to cut API access, they have indicated a willingness to continue to provide OpenAI with access for benchmarking and safety evaluations, which are considered standard practices in the industry. However, details about how this API restriction will affect those evaluations remain unclear.
The events highlight an ongoing competitive tension between AI companies and the regulatory and ethical challenges that come with advancing technology.
