Guide

How to opt out of AI training (every major platform)

Updated July 2026 · Human Data Rights Coalition

Quick answer Updated July 2026

To opt out of AI training, you currently have to adjust settings on each platform separately — most opt you in by default. In ChatGPT, Settings → Data Controls turns off training; Google, Meta, LinkedIn, and X each have their own AI/data-use toggle; and website owners can block AI crawlers in robots.txt. There is no single universal opt-out yet — which is exactly the enforceable right the Human Data Rights Coalition campaigns for.

  • Opting out mostly stops future training — it rarely removes data already used.
  • Companies typically opt you in by default; you must opt out manually.
  • Website owners block AI crawlers via robots.txt.

Opt out, platform by platform

Work through the platforms you actually use. Each takes a minute; settings change often, so re-check periodically.

  1. ChatGPT (OpenAI). Log in, open Settings → Data Controls, and turn off "Improve the model for everyone." This stops your future chats from being used for training.
  2. Google (Gemini & search data). In Gemini, open Settings → Gemini Apps Activity and turn it off. Use Google's "My Activity" controls to limit data retention.
  3. Meta (Facebook & Instagram). Go to Settings → Privacy Center and submit the "Data subject rights" / AI objection form to object to your public content being used for AI.
  4. LinkedIn. Open Settings → Data Privacy → "Data for Generative AI Improvement" and toggle it off.
  5. X (Twitter). Open Settings → Privacy and safety → Grok, and disable data sharing for model training.
  6. Anthropic (Claude). Consumer Claude chats are not used for training by default; review Settings → Privacy to confirm your data-use preferences.
  7. Your own website. Add disallow rules for AI crawlers (GPTBot, ClaudeBot, Google-Extended, CCBot, PerplexityBot) to your robots.txt, and use noai/TDM-reservation meta tags where supported.

What opting out can and can't do

Opting out generally prevents your data from being used in future training. It usually does not remove data already baked into a trained model — "machine unlearning" remains an unsolved technical problem. For data a company still stores, you can also file a deletion request under GDPR (right to erasure) or CCPA/CPRA.

Why this is harder than it should be

The fact that protecting your own data requires hunting through a dozen different settings pages — opted in by default at every step — is the problem. The right to opt out the coalition advocates would make this a single, enforceable action instead of a scavenger hunt. If you think one clear opt-out should exist, join the movement.

Turn knowing your rights into having them

Every person who joins makes the case for enforceable data rights stronger. It's free and takes a minute.

Join the Movement

Opting out of AI training: FAQ

Can I fully remove my data from an AI model that already trained on it?
Usually not completely. You can stop future training and request deletion of stored data under GDPR/CCPA, but removing data already embedded in a trained model ("machine unlearning") is an unsolved technical problem. Opting out today mainly prevents future use.
Does opting out of ChatGPT delete my past conversations from the model?
No. Turning off "Improve the model for everyone" in ChatGPT's Data Controls stops your future chats from being used for training, but it does not remove data already used. Use the data-deletion request process for stored data.
How do website owners stop AI from training on their content?
Add disallow rules for AI crawlers to your robots.txt — GPTBot, ClaudeBot, Google-Extended, CCBot, and PerplexityBot — and use the TDM (text-and-data-mining) reservation and noai/noimageai meta tags where supported.