Claude

Consumer Claude chats and Claude Code sessions may be used for model improvement unless the user turns off Help Improve Claude.

Explicit feedback and safety-review workflows can still route related conversations into model-improvement or safeguard-improvement processes.

Turning off model improvement stops new chats and eligible stored chats from future model training.

Anthropic says business-offering content is governed by customer agreements rather than the consumer Claude settings.

Anthropic says it does not sell personal data as defined by applicable law.

Targeted advertising sharing controls are available through Privacy Choices or global privacy controls.

Sharing categories include affiliates, legal/safety cases, corporate transactions, and user-directed disclosures.

Service-provider and business-partner sharing is disclosed for operational and research purposes.

Marketing cookies are used to promote Anthropic products and measure marketing.

Browser Do Not Track signals are not honored.

Privacy Choices and global privacy controls are supported for advertising-related sharing/use.

Analytics cookies are used to understand and improve service usage.

If model improvement is allowed, data may remain in de-identified training pipelines for up to 5 years.

Safety-flagged content and safety classification data can have longer retention windows than normal deleted chats.

Feedback can keep the related conversation for longer review and improvement workflows.

Deleted consumer chats are removed from history immediately and deleted from backend systems within 30 days.

Model-improvement settings are self-serve in Privacy settings.

Export is available for active Free, Pro, and Max accounts on web or desktop.

Users can delete conversations and delete the account.

Mobile export is not supported, and some deletion cases require support.