Debate erupts over Codex data safety: are your chats really private?
AaronBergman18 · x · 2026-09-08
Reacting to @sholtodouglas's argument that it is "extremely unlikely" user data influenced a recent incident and that Codex user data "surely is" safe, Aaron Bergman pushes back: your data isn't literally safe with Codex (or Anthropic), since chats can end up in front of human reviewers without consent. He adds that OpenAI is the lowest-integrity company he knows of, so such a breach wouldn't shock him — though he'd put the odds below 50%.
The exchange is an outside-view dispute over AI vendors' user-privacy and training-data boundaries, with no hard evidence on either side.
Related event: Developer Denies Claims That Codex Trained on User Data(4 posts)→
More from Models
- Blogger's internal benchmarks: Chinese AI models pricier than US models quality-adjusted — kevinnbass · 2026-09-08
- Zvi: OpenAI's Astra Marks a Rapid Decline in Chain-of-Thought Monitorability — Don't Worry About the Vase (Zvi) · 2026-09-08
- One confidence signal controls LLM abstention: 66.5% to 7% in Gemma 3 27B — dejanseo · 2026-09-08
- NVIDIA details how Domyn specializes Nemotron models for regulated industries — NVIDIAAI · 2026-09-08
- GPT-6 Beats All 48 Levels of Neal.fun's 'I'm Not a Robot' Game as CAPTCHAs Crumble — APPSO · 2026-09-08
- V4.1 Flash is ~3x faster than V4-Flash, smarter per token and image-hungry — teortaxesTex · 2026-09-08