Nvidia, Palantir, and others are reported to restrict the use of OpenAI and Anthropic's most advanced AI models.

On September 14, it was reported that Nvidia, Palantir, and Booz Allen Hamilton are demanding new security guarantees from OpenAI and Anthropic, and are reducing or halting their use of the two companies' most advanced AI models. Palantir has reportedly persuaded Anthropic to commit to "irrevocable" zero data retention guarantees for all models. Some enterprises are turning to in-house AI models to handle more sensitive business tasks.