On August 28, 2025, Anthropic published updates to its Consumer Terms and Privacy Policy. The company is giving users of Claude Free, Pro and Max plans the option to allow their new or resumed chats and coding sessions to be used to improve future Claude models. New users can set this preference during signup; existing users will receive an in-app pop-up and can choose their setting there.
Who is affected and who is not
The changes apply to Claude Free, Pro and Max accounts, including when Claude Code is used from those accounts. The updates do not apply to services governed by Anthropic’s Commercial Terms, such as Claude for Work (including Team and Enterprise plans), Claude for Government, Claude for Education, or API usage — including through third parties like Amazon Bedrock and Google Cloud’s Vertex AI.
Timing and how to choose
Notifications about the policy updates began rolling out on August 28, 2025. Existing users have until October 8, 2025 to accept the updated Consumer Terms and make their choice about model training. If a user accepts now, the choice takes effect immediately and applies only to new or resumed chats and coding sessions. After October 8, selecting a preference for the model-training setting will be required to continue using Claude. The selection can be changed at any time in Privacy Settings.
Data retention: five years or 30 days
If a user opts in to allow their data to be used for model training, Anthropic will retain the eligible new or resumed chats and coding sessions for up to five years to support model development and safety improvements. If a user does not opt in, the existing 30-day retention period remains in force. The five-year retention also applies to feedback submitted about Claude’s responses.
Anthropic notes that deleted conversations will not be used for future model training. If a user withdraws consent or deletes their account, their data will be excluded from future model training; however, data already incorporated into training runs or models in progress may remain in those processes.
Rationale
Anthropic explains that large language models benefit from real-world interaction data because those interactions provide signals about useful and accurate responses. Examples cited include developer-AI coding collaborations, which can provide valuable training signals for future coding tasks. Longer retention helps maintain data consistency across multi-year development cycles (the company notes many models begin development 18–24 months before release) and improves classifiers used to detect misuse such as abuse, spam or other harmful patterns.
Privacy protections and restrictions
To protect user privacy, Anthropic says it uses automated processes and a mix of tools to filter or obfuscate sensitive data. The company also states that it does not sell users’ data to third parties.
What users need to do
- New users will see the model-training preference during signup.
- Existing users will receive an in-app notification and can choose immediately or select “not now” and decide later; the deadline to choose is October 8, 2025.
- The five-year retention policy applies immediately to new and resumed chats and coding sessions if a user opts in.
- Users can change their selection at any time in Privacy Settings; turning off the setting stops new chats from being included in future training, but previously started training runs that used the data will not be retroactively changed.
The update is positioned as a way to improve Claude’s capabilities and safety while preserving user choice over whether their data contributes to future model development.
