We wish to inform you that Anthropic (developer of Claude AI tool) has announced significant updates to its Consumer Terms and Privacy Policy, which took effect on August 28, 2025.
Among other changes, Anthropic may now use, by default, users’ data from certain services and plans for training and improving its AI models. If users do not want their data, coding sessions, and chats to be used for these purposes, they must actively opt out by selecting the appropriate option in a pop-up window that will be presented to existing users, or by changing their default Privacy Settings. Existing users must make their data sharing selection by September 28, 2025.
These updates apply only to Claude Free, Pro, and Max plans (including Claude Code usage from these accounts). They do not affect services under Commercial Terms, including Claude for Work, which includes Anthropic’s Team and Enterprise plans, Anthropic API, Amazon Bedrock, or Google Cloud’s Vertex API, Claude Gov and Claude for Education.
KEY CHANGES:
- Optional Data Sharing: Users can choose whether their interactions with Claude will be used to train and improve Claude’s AI models. This setting is controllable and can be changed at any time in the Privacy Settings. New users may indicate their preferences during the registration process. Existing users will be presented with the option to select their preferences via a pop-up window. Existing users must make their data sharing selection by September 28, 2025 to continue using Claude. These updates will apply only to new or resumed chats and coding sessions.
- Extended Data Retention: For users who allow Anthropic to use their data for model training, Anthropic is extending data retention from 30 days to five years. This applies only to new or resumed conversations. Users who do not allow Anthropic to use their data for model training will maintain the standard 30-day retention period.
A common misconception among users is that subscription-based AI services automatically protect input data from being used in AI model training. However, as stated above, this is not always the case, and in some instances, proactive measures must be taken to prevent such inputs from being used for unintended purposes. Failure to take these steps can result in the unintentional exposure of confidential or proprietary data, unintended use by the service provider, including their incorporation into AI models, potentially breaching confidentiality agreements, compromising intellectual property, infringing privacy rights, and violating data protection laws.
To minimize risk and ensure compliance, organizations and individuals using Claude’s services should exercise heightened caution regarding these new data usage practices. They should regularly review relevant agreements and privacy settings, monitor updates published by AI companies, and ensure that data is not used for training without explicit consent.
Moreover, these amendments to Anthropic’s terms underscore the importance of vigilance in managing AI data use. We generally recommend that organizations adopt a comprehensive compliance framework, including relevant procedures, policies, oversight, and audits, to ensure ongoing compliance with applicable laws, contractual obligations and to protect their proprietary assets.
We recommend companies that use the Claude AI tool review how these changes may impact their operations.
For the full announcement see – Updates to Consumer Terms and Privacy Policy
The content in this update is provided for informational purposes only and is not intended to be comprehensive. It does not serve to replace professional legal advice required on a case by case basis.
