Anthropic is asking users of its Claude chatbot to voluntarily share voice conversations to improve its AI models. This initiative expands the company's data-sharing policy, which previously focused on text and coding sessions.
The policy shift began on August 28, 2025, when Anthropic updated its consumer terms for Free, Pro, and Max plans. Previously, the company did not use consumer data for model training, and audio recordings were deleted immediately after transcription.
Users who opt in to data sharing can have their information retained for up to five years. Those who opt out retain the previous policy, where data is kept for approximately 30 days. Existing users were given an opt-in deadline of September 28 or October 8, 2025.
The "Help improve Claude" setting was enabled by default for users. Deleted conversations are excluded from future training, and users can change their privacy settings at any time to reverse their consent.
These changes apply only to consumer plans. Commercial products, including API and enterprise offerings, are excluded from the new data-sharing terms to maintain client trust.
