Anthropic Users Confront New Decision – Decline or Contribute Data for AI Training

Anthropic Announces Certain Claude Models Can Now Conclude Harmful Or Abusive Dialogues Anthropic Announces Certain Claude Models Can Now Conclude Harmful Or Abusive Dialogues

Anthropic is changing how it handles user data. By September 28, all Claude users must choose whether their conversations can train AI models. This is a major shift.

Previously, Anthropic deleted user chat data within 30 days, except flagged content which it kept up to two years. Now, it wants to use all user chats and coding sessions for training unless users opt out. Data retention can extend up to five years for those who don’t opt out.

This policy hits Claude Free, Pro, Max, and Claude Code users. Business customers on Claude Gov, Work, Education, or API access are exempt.

Advertisement

Anthropic says this move helps improve model safety and future Claude skills like coding and reasoning.

“By not opting out, users will help us improve model safety, making our systems for detecting harmful content more accurate and less likely to flag harmless conversations. Users will also help future Claude models improve at skills like coding, analysis, and reasoning, ultimately leading to better models for all users.”

But this change is largely about Anthropic needing more data to compete with OpenAI and Google.

The timing coincides with growing industry pressure. OpenAI faces a court order to keep all ChatGPT conversations indefinitely, even deleted ones, due to lawsuits from The New York Times and publishers.

OpenAI COO Brad Lightcap called the order:

“a sweeping and unnecessary demand” that “fundamentally conflicts with the privacy commitments we have made to our users.”

Anthropic’s update rollout includes a popup with a large “Accept” button and a smaller toggle for data sharing that defaults to on. This design risks users consenting without realizing they allow data use for training.

Privacy experts warn that complex AI policies make real user consent nearly impossible. The FTC has warned against hiding policy changes “behind hyperlinks, in legalese, or in fine print.”

Whether the FTC is actively monitoring now, with only three commissioners left, is unclear. We’ve reached out to the agency for comment.

Image:

Image Credits: Anthropic

Add a Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Advertisement