Anthropic has revoked OpenAI's access to its AI models for allegedly violating terms of service. This comes as OpenAI prepares to launch its GPT-5 model. Anthropic allows API access for benchmarking and safety evaluations but restricts use for building competing products.