Major tech firms threaten to restrict OpenAI and Anthropic models over data privacy concerns

Major technology firms, including Nvidia, Palantir, and Booz Allen Hamilton, are restricting employee use of advanced AI models from OpenAI and Anthropic. The backlash was triggered by a June policy change allowing Anthropic to retain Fable model usage logs for 30 days. Enterprise clients are demanding irrevocable zero-data-retention guarantees to protect their intellectual property. Meanwhile, Microsoft is capitalizing on these privacy concerns by pitching isolated cloud environments and introducing its own strict AI code of conduct.

Enterprise AI model usage restrictions

• ▪Novo Nordisk continues to use Claude but keeps its proprietary data away from the Anthropic model.
• ▪Nvidia limits Anthropic's models to low-risk, less sensitive tasks and routes proprietary internal work to its own Nemotron models.
• ▪Nvidia, Palantir Technologies, and Booz Allen Hamilton have restricted employee use of advanced AI models from Anthropic and OpenAI or threatened to drop them over data privacy concerns.
• ▪Booz Allen Hamilton has barred its employees from using Anthropic's commercial model for proprietary cybersecurity work involving client data.
• ▪Northrop Grumman runs open-source AI models on air-gapped servers to protect its data.

Anthropic 30-day data retention policy

• ▪Microsoft temporarily withheld Anthropic's Claude Fable 5 model from its internal GitHub Copilot model picker in June 2026 while lawyers reviewed the retention terms.
• ▪Anthropic implemented a policy change in June 2026 allowing it to retain usage logs for its Fable model for 30 days to guard against complex attacks.
• ▪Anthropic retains data flagged by its safety classifiers for up to two years, though it maintains that none of this retained data is used for model training.

Corporate data privacy guarantees demanded

• ▪Anthropic and OpenAI state that they do not train models on customer data by default, though both collect anonymized metadata to improve products.
• ▪Anthropic introduced Enterprise Frontier Safeguards on September 1, 2026, allowing business customers to store activity data in their own encrypted cloud storage.
• ▪An unnamed major United States utility abandoned a test of Anthropic's Fable model after Anthropic declined to sign a non-revocable zero-data-retention clause.
• ▪Palantir Technologies has demanded that Anthropic provide irrevocable zero-data-retention guarantees before making Claude models available through Palantir software.

Microsoft competitive positioning

• ▪Microsoft is pitching isolated cloud environments and its own AI offerings to draw clients concerned about data privacy at outside model providers.
• ▪Microsoft posted a provisional code of conduct imposing restrictions on its AI models, including bans on assisting with weapons manufacturing or concealing reasoning traces.

AI development pace slowdown calls

• ▪Anthropic CEO Dario Amodei called for AI companies to slow the pace of advanced model development, warning that capabilities are advancing faster than researchers can control.
• ▪OpenAI CEO Sam Altman and SpaceX CEO Elon Musk publicly endorsed Dario Amodei's call to slow the pace of frontier AI model development.