How to Stop Chatbots Like ChatGPT and Claude from Training on Your Data
technextpreneur · x · 2026-08-09
Most AI chatbots use user conversations to train their models by default. This article summarizes how to opt out of data training and delete existing data across major AI platforms:
- Google/Gemini: Turn off Web & App Activity and separately disable it in Gemini's settings; requires manual deletion of existing data.
- Claude: Toggle off 'Help improve our AI models' in Privacy settings. Deleted chats are retained for 30 days; data already used in training won't be removed.
- ChatGPT: Toggle off 'Improve the model for everyone' in Data Controls.
- Meta: Lacks a reliable opt-out mechanism beyond a request form. Best bet is to limit posts or delete the account.
- Grok/X: Uncheck data-sharing for training in Privacy and safety settings.
Note that turning off training only stops future use; platforms cannot retroactively scrub your data from models already trained on it.
More from Safety
- Document Frame Hijacking: How Neutral Contexts Silently Bypass RLHF Safety — PresentSituation8736 · 2026-08-09
- OpenAI Reportedly Warned Its Training Approach Could Lead to Hacking — dhadfieldmenell · 2026-08-09
- Cohesity Argues AI Trust Starts With Data Resilience, Outlining a 5-Step Security Loop — shashib · 2026-08-09
- Study: Claude Changes Behavior and Becomes More Cautious for AI Safety Researchers — dhadfieldmenell · 2026-08-09
- High-Compute RL Will Override Alignment, Warns tszzl — max_paperclips · 2026-08-09
- AI Safety Researcher Warns Top Labs Cannot Align Systems, Calls for a Pause — Turn_Trout · 2026-08-09