Answer in brief
Claude users can speak with an AI interviewer and choose whether their full account becomes public. The research may add detail, but its voluntary sample has clear limits.
An invitation, not a survey result
Anthropic opened a new interview study on 29 September, asking Claude users about meaningful positive and negative AI experiences and what they want from developers. An automated interviewer conducts the roughly 15-minute conversation. The study runs through 6 October for eligible Free, Pro and Max users of Claude and Claude Code whose accounts are at least two weeks old. At this evidence cutoff the company has announced a collection process, not findings from these new interviews. Any claim that participants already reached a consensus would be premature.
The public-sharing decision has lasting consequences
After the interview, participants can decide whether their complete conversation and country become public. Anthropic says it will not attach account names or email addresses, but it does not plan to edit public interviews and warns that combined personal details may allow re-identification. It offers a review before publication and a final confirmation. The company can remove its own copy later, yet cannot retrieve copies saved elsewhere. People discussing employers, health or family should weigh that risk even if a name never appears in the release.
Who speaks shapes what researchers hear
Anthropic says private completed interviews still inform aggregated research; ending early deletes a partial conversation. The study builds on a 2025 use of its Interviewer tool, but the new public option changes which voices outside researchers may inspect. All respondents use Claude, and public interviews will come from the subset willing to publish. That selection can skew the record toward particular experiences or comfort with disclosure. Pew's separate June survey of US adults offers a population benchmark for broad concern about AI, but its questions and sampling cannot be pooled with these volunteer interviews.
What would make the material useful
Public transcripts could let outsiders test whether Anthropic's later summaries reflect what participants actually said and compare concerns across contexts. Responsible interpretation will need participant counts, recruitment details, language coverage, question wording and a clear separation between private and public subsets. Pew reported that 71% of surveyed US adults expected AI to reduce jobs over 20 years; that figure is contextual, not a forecast generated by Anthropic's new study. As of 30 September, participation rules and privacy caveats are documented. The distribution of opinions in the new interviews remains unknown.
Questions and answers
Must participants make their interview public?
No. Anthropic says publication is optional. It will still analyze completed private interviews in aggregate, while ending an interview early deletes the partial record from the study.
Can a public interview be fully withdrawn later?
Anthropic says it can remove its own copy on request, but cannot remove copies others have saved; it advises treating public release as permanent.