Anthropic has terminated OpenAI’s access to its Claude AI models, following internal misuse of the platform, according to a report by Wired. Sources revealed that OpenAI had been connecting Claude to proprietary tools to benchmark its performance against its own models in areas such as coding, writing, and safety — actions that violated Anthropic’s commercial terms, which prohibit use for developing competing services.
While Anthropic will still permit limited access for benchmarking and safety evaluations, broader API use has been revoked. An OpenAI spokesperson described the practice as industry standard but acknowledged the restriction. Anthropic leadership, including Chief Science Officer Jared Kaplan, previously expressed reluctance in supporting rival companies with Claude access.
