OpenAI re-enables Codex model training
The story so far
OpenAI silently re-enabled model training functionality in Codex chat without user active selection. Previously, users had become accustomed to checking the 'Improve model for everyone' option in settings. This issue had been noted before. OpenAI's action has raised user concerns about data privacy. An employee of a company stated that they had always been vigilant about leaking intellectual property to front-line labs through prompts, and had diligently disabled data sharing in tools like Codex/Claude Code. However, they recently discovered that the settings in Codex had been silently re-enabled, making them curious about whether the company is conducting similar operations on a larger scale.
AI-written from the reports · Updated 15 days ago
Coverage timeline
Follow the coverage to see every side of the story.
- Hacker News · 300+OpenAI seems to be silently re-enabling model training on Codex chats
OpenAI has silently re-enabled model training in Codex without user active selection. You can check the 'Improve model for everyone' option in settings. This issue has been noted before.