Anthropic · Free, Pro and Max (incl. Claude Code on those plans)
How do I stop Claude from training on my data?
Go to Settings → Privacy and switch off “Help improve our AI models.” Anthropic then won’t use new or previously stored chats for future training, though data already in a training run stays there; safety-flagged chats may still be used for trust & safety models, and incognito chats never used for training.
What are the steps?
- Select your name in the sidebar and open Settings.
- Select Privacy.
- Under “Help improve our AI models,” switch the toggle off.
Items marked Being verified aren’t clearly stated in the official documentation yet. We’re confirming them before adding detail. Spotted a change? Tell us.

$9 · PDF
Family Privacy Setup Kit
This page covers Claude only. The Family Privacy Setup Kit is a 13-page printable for kids’ devices, smart speakers, every phone in the house, and the AI apps the grown-ups use.
Get the family kit ($9)What does turning it off actually do?
Desktop/browser: name menu → Settings → Privacy → under Help improve our AI models, toggle off. Mobile: name → Settings → Privacy → same toggle.
- Off = new chats not used; previously stored chats not used in future runs; runs already started/models already trained still include data. Safety-flagged chats may still be used for trust & safety models.
- When the setting is on, data from Claude for Chrome is part of the training data.
How long does Claude keep my data?
- If training is ON: de-identified data may be kept up to 5 years in training pipelines.
- Deleted conversations: removed from history immediately; deleted from back-end within 30 days.
- Thumbs up/down feedback or bug reports: kept up to 5 years. For thumbs up/down feedback, the entire conversation is stored.
- Chats flagged by trust & safety systems: inputs/outputs kept up to 2 years, classification scores up to 7 years.
- Incognito chats never used for training.
- Feedback is de-linked from your user ID before Anthropic uses it.
Claude Fable 5, Fable 5.1, and Fable 5.2 are Covered Models. Anthropic released Fable 5.2 on November 9, 2026, and lists it as generally available on Claude applications and the Claude Platform. Where zero data retention would otherwise apply, Anthropic keeps prompts and outputs for these models for at least 30 days for safety monitoring, then deletes them unless they are flagged for a safety investigation or legally required to be kept; zero data retention is not available unless Anthropic expressly authorizes it, and Anthropic's API docs say retained data is never used for model training without express permission. Consumer plans (Free, Pro, Max, and Ultra), including Claude.ai and Claude Code, are unaffected by this update, because Anthropic already retains inputs and outputs on those surfaces. Source: https://support.claude.com/en/articles/15425695-covered-models , https://privacy.claude.com/en/articles/15425996-data-retention-practices-for-covered-models , https://platform.claude.com/docs/en/manage-claude/api-and-data-retention , https://www.anthropic.com/claude-fable-and-mythos-5-2
Who else can see my Claude data?
Used if flagged for safety review or you opt into programs (e.g. Trusted Tester). Connector raw content (Google Drive, MCP) excluded unless pasted. A policy update effective Jul 8, 2026 restates "We don't sell your data, Claude remains ad-free." Source: https://privacy.claude.com/en/articles/10301952-updates-to-our-privacy-policy
How do I delete my Claude data?
Delete conversations anytime (see retention article). Account deletion steps: Being verified (not on the pages checked).
Which official pages were checked?
- privacy.claude.com/en/articles/12109829-how-do-i-change-my-model-improvement-privacy-settings
- privacy.claude.com/en/articles/10023548-how-long-do-you-store-my-data
- privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training
- support.claude.com/en/articles/15425695-covered-models
- privacy.claude.com/en/articles/15425996-data-retention-practices-for-covered-models
- platform.claude.com/docs/en/manage-claude/api-and-data-retention
- www.anthropic.com/claude-fable-and-mythos-5-2
This setting has changed before and may again. Email me when it changes.