Anthropic, the company behind Claude, has made an announcement that indicates a move from its existing privacy-first approach.

It will train its AI models on transcripts and sessions from user accounts, with data to be retained for up to 5 years, unless users opt out. More here from Perplexity.