Anthropic's Clio Privacy System at Center of Debate Over Opt-Out Data Gaps

austinc3301 · x · 2026-09-13

A debate over how AI providers derive insights from user chats. Anthropic's approach relies on Clio (now Anthropic Insights), a privacy-preserving system that analyzes real-world chat logs in aggregate — analogous to Google Trends — while claiming rigorous privacy protection and feeding into safety improvements.

The core criticism: opt-out mechanisms may only cover literal chat contents, leaving other insight-extraction pathways from opted-out data unspecified. OpenAI was reportedly asked these questions thousands of times and strategically avoided answering, with no formal response to date.

Original post →

More from Safety

Safety channel →