AI labs still have a data trust problem

The piece says AI firms’ promises not to train on customer data haven’t resolved trust issues: Anthropic’s decision to retain Fable usage logs for 30 days prompted Palantir, Nvidia and Booz Allen Hamilton to pull back on sensitive work.

• Nvidia and Others Restrict Anthropic’s Fable Over Data Retention2026-09-16
• Nvidia: robots await a ChatGPT moment2026-09-16
• Security experts urge basics before external AI audits2026-09-17
• CNBC on concerns about proposed AI risk evaluators2026-09-16
• Lawyer says Anthropic and OpenAI leaders 'untrustworthy'2026-09-16