Anthropic Changes Its Data Policies: Opt Out or Share for AI Training
Anthropic has changed its data policies, giving users the choice between opting out or sharing their data to improve AI models, with a decision deadline in September 2025.
Anthropic's data policy overhaul
Anthropic has announced a major change in how user data is handled, introducing a mandatory choice for users of its artificial intelligence platforms. From a cutoff date, users must decide whether to share their conversation data for training AI models or keep the previous policy of deleting data within 30 days.
The new data management model
Until now, Anthropic did not use users' chat data to improve its AI models. The new policy, instead, provides for the use of conversations and coding sessions of users who choose to share their data, with an extension of active retention to up to five years.
Conversations and coding content will be collected and used to refine the safety and comprehension capabilities of Anthropic's AI models, improving accuracy in detecting harmful content and the effectiveness of responses.
Implications for users
Users of products such as Claude Free, Pro and Max must express their preference through a prompt during the sign-up process or through a policy update pop-up window, with a default acceptance that can be turned off, but with the risk of unintended outcomes.
The decision must be made by September 28, 2025: by accepting, you allow data to be collected and used for up to five years; by declining, the previous 30-day deletion policy is kept.
Broader context and ethical considerations
Anthropic's new policies fit into a broader picture of tension between AI companies and privacy regulations, with giants like OpenAI under legal scrutiny for their data handling methods. Alongside these dynamics, there is a growing risk of uninformed choices by users, often influenced by interface design that underestimates the complexity of privacy decisions.
Future prospects and regulatory challenges
The extension of data retention reflects an industry trend toward preferring long-lived data to improve model quality and increase safety, even though it raises the risks of exposure and privacy violations. Regulators, such as the FTC, are closely monitoring these practices, highlighting the need for transparency and truly informed consent.
Conclusion and practical aspects
- AI companies are shifting the balance between innovation and privacy through more flexible opt-in policies, but with risks of automatic approval that can harm user awareness.
- Users must pay attention to the new settings and decisions, considering that long-term data collection can affect future privacy and the use of their data in AI models.
- Transparency in data collection and usage practices will be one of the key factors for consumer trust and regulatory compliance.