OpenAI Reportedly Re-Enabling User Data Training Settings Automatically
A Hacker News user reported that OpenAI periodically re-enables the setting allowing user chat data to be used for model training, even after users manually opt out. The poster noted they carefully recorded when they toggled the setting off, only to find it switched back on upon checking later. If user opt-out settings are silently reset, it undermines trust and raises serious privacy concerns regarding consent and user data control in AI services. It highlights potential dark patterns or technical flaws in how AI providers handle privacy preferences. Community members noted that simply toggling the setting in the ChatGPT interface may not be sufficient, recommending that users also submit a formal request via OpenAI's privacy portal. Additionally, location-based regulatory protections like the EU's GDPR may influence whether users experience this setting reset.
## BACKGROUND
Generative AI providers frequently rely on user prompts and conversations to train and improve future iterations of their models. To address privacy concerns, companies typically offer opt-out toggles in account settings or privacy request forms so users can prevent their data from being ingested into training datasets.