Anthropic is asking users of its Claude chatbot to voluntarily share voice conversations to improve its AI models. This request follows a major policy change on August 28, 2025, when the company updated its consumer terms of service to allow training on user data. By default, the ‘Help improve Claude’ toggle was enabled, with users who opt in having their data retained for up to five years, compared to approximately 30 days for those who opt out. Old conversations are not swept in retroactively and deleted conversations are excluded from future training runs. This change only affects consumer offerings (Free, Pro, and Max), with commercial products such as the API and enterprise offerings not being affected.
Source: Read the original article
