OpenAI has been reportedly disconnected from Anthropic’s APIs following a violation of the terms of service. As per Wired, OpenAI was utilizing Anthropic’s Claude Code to aid in the development of GPT-5, which is scheduled for release in August. Sources indicate that OpenAI used Claude’s internal tools to evaluate GPT-5’s efficacy in coding and creative writing, alongside safety prompts associated with CSAM, self-harm, and defamation, breaching Anthropic’s commercial terms that forbid using its tools to create rival AI products. OpenAI remarked that testing models against competitors is typical within the industry and expressed frustration over the API access being rescinded, highlighting that Anthropic still retains access to OpenAI’s API. A representative noted that OpenAI’s access would be restored for benchmarking and safety assessments. Previously, Anthropic also revoked Windsurf’s API access amidst speculation of its sale to OpenAI, which did not occur. Moreover, Anthropic has modified its rate limits for Claude, effective late August, due to certain users contravening policies by sharing and reselling accounts.
Disclosure: Ziff Davis, the parent company of Mashable, initiated a lawsuit against OpenAI in April, claiming copyright infringement in the training and operation of its AI systems.