Metraipolis › archive › dc_390eb3bfeef3

Measuring Personal Information Exposure in Mental Health-Related Conversations with LLMs — arXiv:2610.41045

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.41045
License (as archived)
CC0-1.0
Archived
2026-11-03T05:39:56.724Z
SHA-256 of the archived text
3a77ce445cf4abe216e2b5f65eaad7c574e035fc60d2157ee192408f8fd9436d
Stable URL of this page
https://metraipolis.com/documento/dc_390eb3bfeef3

How to cite

Measuring Personal Information Exposure in Mental Health-Related Conversations with LLMs — arXiv:2610.41045 — arXiv (en), https://arxiv.org/abs/2610.41045 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_390eb3bfeef3 (Metraipolis archive dc_390eb3bfeef3, sha256 3a77ce445cf4abe216e2b5f65eaad7c574e035fc60d2157ee192408f8fd9436d)
BibTeX
@misc{dc390eb3bfeef3,
  title = {Measuring Personal Information Exposure in Mental Health-Related Conversations with LLMs — arXiv:2610.41045},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41045},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_390eb3bfeef3; sha256 3a77ce445cf4abe216e2b5f65eaad7c574e035fc60d2157ee192408f8fd9436d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41045. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) are increasingly used for informal mental health support, yet the privacy implications of conversational interaction remain poorly understood. Prior LLM privacy research has focused largely on model-side risks such as memorization and adversarial extraction, leaving personal information inferred from ordinary conversation understudied. We introduce a profile-recovery framework that measures personal information exposure by comparing user profiles with attributes inferred from conversation alone. We generate 240 synthetic multi-turn mental health conversations from profiles sampled from a nationally representative health survey and validate them against real counseling transcripts, finding comparable overall disclosure volume. Across 21 attributes, 33-41% of each profile is recoverable, with 76.5% first-inference and 94.3% any-of-3 accuracy. 239 of 240 conversations expose at least one sensitive mental health attribute, and 45.0\% expose all six. Our findings reveal a gap between disclosure and inferability, motivating HCI approaches to make inference risks legible and support privacy-aware interaction design.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.HC, nuovi lavori del giorno).
Categorie arXiv: Human-Computer Interaction (cs.HC).
Pagina del lavoro: https://arxiv.org/abs/2610.41045 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T05:23:57.962Z dalla pagina https://arxiv.org/list/cs.HC/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).