Understanding Privacy Norms Around LLM-Based Chatbots: A Contextual Integrity Perspective
LLM-driven chatbots like ChatGPT have created unprecedented volumes of conversational data, yet little is known about user privacy expectations for this information. We surveyed 300 US ChatGPT users to understand privacy norms around chatbot data sharing using the contextual integrity framework. Our findings reveal a stark disconnect between user concerns and behavior. While 82% of respondents rated chatbot conversations as sensitive or highly sensitive—more than email or social media posts—nearly half reported discussing health topics and over one-third discussed personal finances with ChatGPT. Participants expressed strong privacy concerns (t(299) = 8.5, p
Authors
- Bill Howe (ORCID: https://orcid.org/0000-0001-8588-8472)
- Sarah Tran
- Yue Fu (ORCID: https://orcid.org/0000-0001-5828-5932)
- Hongen Lu
- Robert Wolfe (ORCID: https://orcid.org/0000-0001-7133-695X)
- Alexis Hiniker (ORCID: https://orcid.org/0000-0003-1607-0778)
- Isaac Slaughter (ORCID: https://orcid.org/0000-0002-1911-2374)
- Bernease Herman (ORCID: https://orcid.org/0000-0002-5453-4994)
- Nicholas Weber (ORCID: https://orcid.org/0000-0002-6008-3763)
- Lufei Chen
- Biniyam Gebreyohannes
Institutions
- University of Washington (US)
- Rutgers Sexual and Reproductive Health and Rights (NL)
Publication Details
- Journal
- Proceedings of the AAAI/ACM Conference on AI Ethics and Society
- Published
- 2025-10-15
- DOI
- https://doi.org/10.1609/aies.v8i3.36735
- Citations
- 1
- Primary Topic
- Privacy-Preserving Technologies in Data
- Type
- article
- Field-Weighted Citation Impact
- 2.48
Funders
- Alfred P. Sloan Foundation