Anthropic Revokes OpenAI API Access Over GPT-5 Use

Anthropic has revoked OpenAI's API access to its Claude AI models, citing violations of terms of service after OpenAI's technical staff used special developer access for benchmarking and internal testing to aid GPT-5 development. Anthropic claims this crossed into prohibited territory of building competing AI models, although it acknowledges benchmarking is standard and will continue to allow API access for certain safety and benchmarking purposes. OpenAI maintains its actions are industry standard and expressed disappointment, pointing out that its own API remains accessible to Anthropic. The dispute underscores increasing competition and sensitivities among leading AI companies as they protect intellectual property. This move follows similar API access restrictions previously imposed by Anthropic on other potential competitors. The conflict unfolds just ahead of GPT-5's anticipated launch.