Palantir, Nvidia pulled back from Anthropic over 30-day log retention, exposing AI labs' data trust problem

The Decoder · rss · 2026-09-16

The Decoder reports that AI labs have a data trust problem their policies haven't solved. OpenAI and Anthropic promise corporate customers their data won't train models, yet when Anthropic said it would store usage logs of flagship model Fable for 30 days, Palantir, Nvidia, and Booz Allen Hamilton pulled back from using it for sensitive work. Verbal privacy commitments, the article argues, still can't replace technical guarantees.

Original post →

More from Companies & People

Companies & People channel →