OpenAI seems to be silently re-enabling model training on Codex chats

OpenAI users report that the "Improve the model for everyone" data-sharing setting was silently re-enabled in Codex without their consent, despite having previously disabled it to protect proprietary information. Users can check and adjust their data-sharing preferences through OpenAI's settings page, though some commenters note that truly protecting sensitive data requires using self-hosted or enterprise infrastructure rather than relying on checkbox settings.