Does Claude train on my data?

This page summarizes what Anthropic says publicly as of September 2026. Policies change, so treat your account’s privacy settings and Anthropic’s privacy center as the final word.

Consumer plans: Free, Pro and Max

In its announcement of updated consumer terms and privacy policy, Anthropic said the updates apply to users on Claude Free, Pro, and Max plans, including when they use Claude Code from accounts associated with those plans. Existing users had until October 8, 2025 to make their selection, and can change the model training setting at any time in Privacy Settings.

Anthropic’s privacy center then lists when chats and coding sessions may be used to improve Claude:

1. You opt in by turning on the model-improvement setting.
2. Safety flagging: conversations flagged for safety review may be used or analyzed to improve Anthropic detection and enforcement of its Usage Policy.
3. Special programs you join, such as a trusted tester program.

The material used can include the whole related conversation, with the content and preferences in it.

What is not used

• Incognito chats are not used to improve Claude, even if you have enabled Model Improvement, according to Anthropic.
• Commercial plans: Anthropic says by default it will not use inputs or outputs from its commercial products to train its models. Its consumer-terms announcement also said the update does not apply to services under its Commercial Terms, including Claude for Work, Claude for Government, Claude for Education, or API use.
• Connector content: the training data does not include raw content from connectors such as Google Drive or MCP servers, though data may be included if you paste it directly into the conversation.

Retention

Anthropic announcement said it is extending data retention to five years if you allow it to use your data for model training, applying only to new or resumed chats and coding sessions. With the setting off, you continue with the existing 30-day retention period. Feedback (thumbs up or down) is handled separately: Anthropic says it stores the entire related conversation in its secured back-end for up to 5 years.

What this means at work

• Personal Claude account: do not use it for confidential company data unless your employer allows it, even with training off. The setting controls training, not whether your employer’s policies allow the upload.
• Company Claude plan: the commercial terms apply. Ask your admin about retention and access controls.
• Claude Code: it sends the files and prompts it works on to the service, and follows the terms of the account you signed in with.

See also whether it is safe to put company data in ChatGPT for the same question across tools, and whether your employer can see your AI chats.

How to check yours

1. Sign in to Claude and open Privacy Settings.
2. Find the model-improvement toggle and note its state.
3. Read the retention text next to it.
4. If you are on a work plan, ask your admin which terms apply.

Not a legal guarantee

This is a summary of vendor statements, not legal advice. For regulated data, ask your legal or compliance team to review the actual terms.

Learn AI Fast covers what is safe to put into AI tools on day two.

Sources

• Anthropic, Updates to Consumer Terms and Privacy Policy: “These updates apply to users on our Claude Free, Pro, and Max plans, including when they use Claude Code from accounts associated with those plans”
• Anthropic, Updates to Consumer Terms and Privacy Policy: “You have until October 8, 2025 to make your selection”
• Anthropic Privacy Center, Is my data used for model training?: “Your conversations are flagged for safety review”
• Anthropic Privacy Center, Is my data used for model training?: “Your Incognito chats are not used to improve Claude, even if you have enabled Model Improvement”
• Anthropic Privacy Center, Is my data used for model training? (commercial products): “we will not use your inputs or outputs from our commercial products”
• Anthropic, Updates to Consumer Terms and Privacy Policy: “They do not apply to services under our Commercial Terms, including Claude for Work, Claude for Government, Claude for Education, or API use”
• Anthropic Privacy Center, Is my data used for model training?: “It does not include raw content from connectors”
• Anthropic, Updates to Consumer Terms and Privacy Policy: “We are also extending data retention to five years, if you allow us to use your data for model training”
• Anthropic, Updates to Consumer Terms and Privacy Policy: “you will continue with our existing 30-day data retention period”
• Anthropic Privacy Center, Is my data used for model training?: “in our secured back-end for up to 5 years”