Palantir, Nvidia and Booz Allen Hamilton are limiting advanced AI models over concerns that Anthropic and OpenAI could retain or derive value from sensitive corporate information. Palantir has sought zero-data-retention guarantees from Anthropic, Nvidia confines Anthropic models to less sensitive work while favoring its Nemotron models for proprietary tasks, and Booz Allen has barred Anthropic's commercial model from proprietary cybersecurity projects. Anthropic's 30-day retention policy, introduced with Fable 5 in June, prompted objections from enterprise customers despite the company's statement that retained data is not used for training. Its Enterprise Frontier Safeguards will allow customers to store monitoring data in their own encrypted cloud infrastructure without Anthropic employee review. Nvidia shares fell roughly 3% on Monday as AI-related stocks broadly declined.