On Tuesday, Anthropic revoked OpenAI’s access to its AI models for allegedly violating terms of service, as reported by Wired. This decision comes as OpenAI, led by Sam Altman, is poised to release its new GPT-5 model. Anthropic’s spokesperson, Christopher Nulty, stated that OpenAI’s technical team used Claude tools improperly, confirming this action is a “direct violation” of their service agreements. Although access has been cut off, Nulty indicated that Anthropic would still provide OpenAI API access for benchmarking and safety evaluations. Reports suggest OpenAI had utilized Claude through special developer APIs to test its capabilities, including coding tasks and responses to sensitive prompts like CSAM and self-harm. OpenAI’s Chief Communications Officer, Hannah Wong, expressed disappointment over the cutoff but reiterated that evaluating AI systems is an industry standard. This isn’t the first instance of Anthropic restricting competitor access, as seen with Windsurf earlier this year.
Source link
Share
Read more