Anthropic is making significant changes to its handling of user data, requiring all Claude users to decide by September 28th whether they allow their conversations to be used to train AI models.
Anthropic is making significant changes to its handling of user data, requiring all Claude users to decide by September 28th whether they allow their conversations to be used to train AI models.
Anthropic previously didn’t use user chat data to train its models. Now, the company wants to train its AI systems on user conversations and coding sessions, and has said that for those who don’t opt out, the data retention period will be extended to 5 years, TechCrunch writes .
This is a major update. Previously, users of Anthropic consumer products were informed that their queries and conversation results would be automatically deleted from Anthropic’s backend within 30 days, “unless required by law or policy to retain them for longer,” or if the input was flagged as violating policy, in which case user input and output data could be retained for up to 2 years.
The new rules apply to Claude Free, Pro, and Max users, including those using Claude Code. But they do not apply to business customers using Claude Gov, Claude for Work, Claude for Education, or the API.
Anthropic says that by opting out, users will “help us improve the model’s security by making our systems for detecting harmful content more accurate and less likely to flag harmless conversations.” Users will also “help future Claude models improve skills like coding, analysis, and reasoning, ultimately leading to better models for all users.”
But the full truth is probably a little less selfless.
Like any other large AI model development company, Anthropic needs more data. Training AI models requires vast amounts of high-quality conversational data, and Claude’s access to millions of interactions should provide the kind of real-world content that can improve Anthropic’s competitive position.
