Giving chat feedback gives Anthropic your data

Anthropic does not use inputs or outputs from its commercial products for model training by default, but explicitly providing feedback through thumbs up/down buttons allows the company to store entire conversations for up to five years and potentially use them to train AI models. Users can disable feedback submission through organization settings, and Anthropic de-links feedback from user identities before using it for research and model improvement. Raw content from external connectors is excluded from feedback storage, though directly copied content into conversations may be retained.