Anthropic has revoked OpenAI’s access to the Claude Code application-programming interface after internal monitoring tools flagged sustained use of the coding model by OpenAI engineers. Anthropic said the activity breached a clause in its terms of service that prohibits customers from using Claude to train, test or reverse-engineer competing artificial-intelligence systems. The suspension comes days before OpenAI is expected to unveil GPT-5, its next large language model. While full development access has been cut, Anthropic said it will continue offering limited API availability to OpenAI for benchmarking and safety evaluations, maintaining what it called “industry-standard” collaboration on risk assessments. The clash underscores intensifying rivalry in the commercial AI market. Claude has gained popularity among corporate developers, and research firm Menlo Ventures recently estimated Anthropic’s share of enterprise AI-API spending at 32 percent, surpassing OpenAI’s. Neither company disclosed how long the restriction may remain in place.