Anthropic Revokes OpenAI Access to Claude Over ToS Breach Ahead of GPT-5 Launch

Anthropic has revoked OpenAI's access to its Claude API, alleging that OpenAI violated the terms of service by using Claude's coding tools to develop and benchmark its upcoming GPT-5 model. Anthropic's terms explicitly prohibit using the service to build competing AI models or reverse engineer the technology. OpenAI admitted to benchmarking Claude extensively within its internal tools for coding, creative writing, and safety-related content, defending this as industry-standard practice to improve AI performance and safety. Despite respect for Anthropic's decision, OpenAI expressed disappointment given that Anthropic's API access remains open to them. Anthropic stated it will continue to allow API access for benchmarking and safety evaluations, although it did not clarify how this would apply following the ban. This dispute highlights growing tensions among leading AI companies in the weeks before the critical GPT-5 launch.