How to Stop Claude (Anthropic) from Training on Your Data in 2026
At a glance
- Platform Anthropic Claude
- Company Anthropic
- Grace period Training controls and account deletion are separate. Review Anthropic's current retention policy for the data you want deleted.
- Official page privacy.claude.com
What you need to know first
Separate actions: opt out of model training (Privacy Settings toggle), use Incognito mode (never trained, not saved), delete individual chats or your whole account.
Control training and delete chats or your account
- Step 1: Sign in at claude.ai (Free, Pro, or Max)
- Step 2: Click your profile/account icon > Settings
- Step 3: Open Privacy Settings
- Step 4: Find 'Help improve Claude' (model-training toggle)
- Step 5: Toggle it OFF and save
- Step 6: To delete a chat on the web, open its sidebar menu, select Delete, and confirm
- Step 7: To delete your account, export anything you need, then open Settings > Account > Delete account. This is permanent
- Step 8: Pro and Max subscribers must cancel their subscription and wait for its current period to end before deleting the account
On mobile
- Open Claude app > profile/menu > Settings
- Open Privacy Settings
- Turn off 'Help improve Claude'
Opt out of AI training
In Claude, open Settings > Privacy Settings and turn off 'Help improve Claude'. Anthropic asks consumer users to choose whether to allow model improvement; check your saved preference rather than assuming a default. Turning this off does not delete your chats or undo completed model training.
Official instructions: delete a chat or delete your account. Sources checked September 26, 2026.
Separate data request
Anthropic Privacy Center (privacy.claude.com) for access/deletion/portability; EU users have full GDPR rights
Key gotchas
- Check your saved model-improvement preference; consumer users are offered a choice
- Opening an OLD conversation after the policy change can make it eligible for training
- Training-on accounts have 5-year retention (vs 30 days opted out)
- Safety-flagged conversations are an opt-out exception
What this doesn't cover
Anthropic does not 'unlearn' data already incorporated into a model; the safety-flag carve-out means flagged chats may still be used regardless of your setting; broker-held data is separate.
Opting out of Claude training limits future model exposure, but can't remove your personal data from the broker ecosystem that aggregates and sells it.
See what personal data is exposed about you beyond Anthropic Claude.
Run a free scan →Frequently asked questions
Does Anthropic Claude delete my data immediately?
Training controls and account deletion are separate. Review Anthropic's current retention policy for the data you want deleted.
Can I opt out of Anthropic Claude using my data for training?
In Claude, open Settings > Privacy Settings and turn off 'Help improve Claude'. Anthropic asks consumer users to choose whether to allow model improvement; check your saved preference rather than assuming a default. Turning this off does not delete your chats or undo completed model training.
Does deleting my Anthropic Claude account remove my data from the internet?
Anthropic does not 'unlearn' data already incorporated into a model; the safety-flag carve-out means flagged chats may still be used regardless of your setting; broker-held data is separate.
What are common mistakes when deleting Anthropic Claude data?
Check your saved model-improvement preference; consumer users are offered a choice. Opening an OLD conversation after the policy change can make it eligible for training.
Sources
Steps and URLs reflect the platform documentation as of the date above. Platforms may change their settings layout at any time. This guide is informational and not affiliated with Anthropic.