Agentic LLMs as Powerful Deanonymizers: Re-identification of Participants in the Anthropic Interviewer Dataset

TL;DR

The paper investigates privacy risks in releasing rich qualitative data via the Anthropic Interviewer dataset, demonstrating that agentic LLMs with web search can re-identify participants by linking transcripts to publications. Using a two-stage, web-augmented attack focused on scientist interviews, the study re-identifies 6 of 24 transcripts with very high confidence, underscoring a low barrier and low cost for such attacks. The findings highlight serious privacy harms for participants and call for mitigations, stronger data governance, and responsible disclosure practices in the era of powerful LLM agents. Overall, the work cautions the research community about de-anonymization risks in qualitative data and provides concrete directions for safeguarding future data releases.

Abstract

On December 4, 2025, Anthropic released Anthropic Interviewer, an AI tool for running qualitative interviews at scale, along with a public dataset of 1,250 interviews with professionals, including 125 scientists, about their use of AI for research. Focusing on the scientist subset, I show that widely available LLMs with web search and agentic capabilities can link six out of twenty-four interviews to specific scientific works, recovering associated authors and, in some cases, uniquely identifying the interviewees. My contribution is to show that modern LLM-based agents make such re-identification attacks easy and low-effort: off-the-shelf tools can, with a few natural-language prompts, search the web, cross-reference details, and propose likely matches, effectively lowering the technical barrier. Existing safeguards can be bypassed by breaking down the re-identification into benign tasks. I outline the attack at a high level, discuss implications for releasing rich qualitative data in the age of LLM agents, and propose mitigation recommendations and open problems. I have notified Anthropic of my findings.