OpenAI seems to be silently re-enabling model training on Codex chats
The point
This Hacker News post is a user report, not a confirmed OpenAI announcement. The poster says his company had disabled data sharing in Codex, but yesterday he and his coworkers noticed the setting had been reenabled and none of them knowingly opted back in.
Why it matters
For builders, it is a prompt to check your own data controls instead of trusting an old opt out.
Key details
- Check at chatgpt.com/settings/data-controls and look for Improve the model for everyone
- The poster also links an alternative at privacy.openai.com/policies, with the take control option
Original ~2 min → 25 sec here Source: Hacker News, Nov 3, 2026