Researchers claim OpenAI's opt-out of training data is misleading and only covers ChatGPT
PMinervini · x · 2026-09-09
Delip Rao highlights that OpenAI makes opting out of training data usage hard to find, buried behind a privacy link. Worse, following Edoardo Contente's documented steps only excludes data from ChatGPT — with no guarantee it applies to internal models such as reward models. The thread argues the data control setting is effectively deceptive, appearing to offer opt-out while not fully preventing training use.
Related event: OpenAI's Data Opt-Out Criticized as Hidden and Limited(4 posts)→
More from Safety
- BlueDot's Frontier AI Governance Course: 10,000+ Alumni Placed in Policy Roles — AndyMasley · 2026-09-09
- Scoop: Altman Privately Opposes US Government Stake in OpenAI, Reversing Signals — ShakeelHashim · 2026-09-09
- First Take It Down Act conviction lands 15-year sentence for AI-generated abuse material — TechNadu · 2026-09-09
- Researchers slam OpenAI over murky data controls, urge clarification or customer exodus — anshulkundaje · 2026-09-09
- Synthetic Data May Leak Answers, Security Expert Questions — matthew_d_green · 2026-09-09
- OpenAI's 'Do not train on my content' option found not enabled by default — anshulkundaje · 2026-09-09