Privacy Perceptions of Custom GPTs by Users and Creators
Open accessThe DOI matches its Crossref record; 82 of the 82 references it cites were found in a registry or an index.
Abstract
GPTs are customized LLM apps built on OpenAI's large language model.Any individual or organization can use and create GPTs without needing programming skills.However, the rapid proliferation of over three million GPTs has raised signifcant privacy concerns.To explore the privacy perspectives of users and creators, we interviewed 23 GPT users with varying levels of creation experience.Our fndings reveal blurred lines between user and creator roles and their understanding of GPT data fows.Participants raised concerns about data handling during collection, processing, and dissemination, alongside the lack of privacy regulations.Creators also worried about loss of their proprietary knowledge.In response, participants adopted practices like self-censoring input, evaluating GPT actions, and minimizing usage traces.Focusing on the dual role of user-creators, we fnd that expertise and responsibility shape privacy perceptions.Based on these insights, we propose practical recommendations to improve data transparency and platform regulations.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Users often rely on inaccurate mental models of LLM data flows and are vulnerable to dark patterns, which impedes their ability to grasp real privacy risks [22,47].”
From Understanding Users' Privacy Perceptions Towards LLM's RAG-based Memory (2025) · cites this paper as [22]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Using thematic analysis in psychology
Braun, V., & Clarke, V. (2006) · Qualitative Research in Psychology · doi:10.1191/1478088706qp063oa · Cited by 194,414
The Measurement of Observer Agreement for Categorical Data
Landis, J. R., & Koch, G. G. (1977) · Biometrics · doi:10.2307/2529310 · Cited by 81,148 · cited 2×
“After discussing disagreements and re-fning the codebook by adding new codes, they repeated the process with another transcript, achieving an IRR score of 0.81, which indicates substantial agreement [57].”
InfluentialEysenck, H. J. (1985) · Behaviour Research and Therapy · doi:10.1016/0005-7967(85)90087-7 · Cited by 31,975
Lazarus, R. S. (1991) · doi:10.1093/oso/9780195069945.001.0001 · Cited by 12,691
Are We There Yet? Data Saturation in Qualitative Research
Fusch, P., & Ness, L. (2015) · The Qualitative Report · doi:10.46743/2160-3715/2015.2281 · Cited by 4,460
An Extended Privacy Calculus Model for E-Commerce Transactions
Dinev, T., & Hart, P. (2006) · Information Systems Research · doi:10.1287/isre.1060.0080 · Cited by 2,640 · cited 1×
“As such, the privacy calculus theory explains that users weigh the perceived benefts (e.g., discounts) against the risks (e.g., identity theft) when deciding to share personal information [29, 38].”
Information Privacy Research: an Interdisciplinary Review1
Smith, H. J., Dinev, T., & Xu, H. (2011) · MIS Quarterly · doi:10.2307/41409970 · Cited by 2,013
(2011) · IEEE Software · doi:10.1109/ms.2011.67 · Cited by 1,989
Privacy-preserving data mining
Agrawal, R., & Srikant, R. (2000) · Proceedings of the 2000 ACM SIGMOD international conference on Management of data · doi:10.1145/342009.335438 · Cited by 1,719 · cited 1×
“…explored technical methods to safe-guard user privacy while maintaining system performance through privacy-preserving algorithms and system design [4, 51, 61, 67], there is limited understanding from the user’s perspective, such as how willing users are to compromise their privacy in exchange for…”
Solove, D. J. (2006) · University of Pennsylvania Law Review · doi:10.2307/40041279 · Cited by 1,388
Research works printed but not linked: 17
- Irwin Altman. 1974. Privacy: A Conceptual Analysis. In Man-Environment Interactions: Evaluations and Applications: Part 2, D. H. Carson (Ed.). Environmental Design Research Association, Washington, DC, USA, 3–28.
- Benjamin Andow, Samin Yaseer Mahmud, Justin Whitaker, William Enck, Bradley Reaves, Kapil Singh, and Serge Egelman. 2020. Actions speak louder than words:{ Entity-Sensitive} privacy policy and data flow analysis with { PoliCheck}. In 29th USENIX Security Symposium (USENIX Security 20). USENIX Association, Boston, MA, USA, 985–1002.
- Natã M. Barbosa, Zhuohao Zhang, and Yang Wang. 2020. Do Privacy and Security Matter to Everyone? Quantifying and Clustering User-Centric Considerations About Smart Home Device Adoption. In Sixteenth Symposium on Usable Privacy and Security (SOUPS 2020). USENIX Association, Virtual Event, 417–435. https://www.usenix.org/conference/soups2020/presentation/barbosa
- Nicholas Carlini, Florian Tramèr, Eric Wallace, Matthew Jagielski, Ariel Herbert-Voss, Katherine Lee, Adam Roberts, Tom Brown, Dawn Song, Úlfar Erlingsson, Alina Oprea, and Colin Raffel. 2021. Extracting Training Data from Large Language Models. In 30th USENIX Security Symposium (USENIX Security 21). USENIX Association, Virtual Event, 2633–2650. https://www.usenix.org/conference/usenixsecurity21/presentation/carlini-extracting
- Jessica Colnago, Lorrie Faith Cranor, Alessandro Acquisti, and Kate Hazel Stanton. 2022. Is it a concern or a preference? An investigation into the ability of privacy scales to capture and distinguish granular privacy constructs. In Eighteenth Symposium on Usable Privacy and Security (SOUPS 2022). USENIX Association, Boston, MA, USA, 331–346.
- Kevin Gallagher, Sameer Patil, and Nasir Memon. 2017. New Me: Understanding Expert and Non-Expert Perceptions and Usage of the Tor Anonymity Network. In Thirteenth Symposium on Usable Privacy and Security (SOUPS 2017). USENIX Association, Santa Clara, CA, 385–398. https://www.usenix.org/conference/soups2017/technical-sessions/presentation/gallagher
- Hamza Harkous, Kassem Fawaz, Kang G Shin, and Karl Aberer. 2016. PriBots: Conversational privacy with chatbots. In Twelfth Symposium on Usable Privacy and Security (SOUPS 2016). USENIX Association, Denver, CO, USA, 6 pages.
- Evin Jaff Yuhao Wu Ning Zhang and Umar Iqbal. 2024. Data Exposure from LLM Apps: An In-depth Investigation of OpenAI’s GPTs. arxiv:https://arXiv.org/abs/2408.13247https://arxiv.org/abs/2408.13247
- Ruogu Kang, Laura Dabbish, Nathaniel Fruchter, and Sara Kiesler. 2015. “My Data Just Goes Everywhere:” User Mental Models of the Internet and Implications for Privacy and Security. In Eleventh Symposium On Usable Privacy and Security (SOUPS 2015). USENIX Association, Ottawa, 39–52. https://www.usenix.org/conference/soups2015/proceedings/presentation/kang
- Amit Kulkarni. 2021. GitHub Copilot AI is Leaking Functional API Keys. Analytics Drift. https://analyticsdrift.com/github-copilot-ai-is-leaking-functional-api-keys/ (Accessed on 08/26/2024).
- Ambar Murillo, Andreas Kramm, Sebastian Schnorf, and Alexander De Luca. 2018. " If I press delete, it’s gone"-User Understanding of Online Data Deletion and Expiration. In Fourteenth Symposium on Usable Privacy and Security (SOUPS 2018). USENIX Association, Baltimore, MD, USA, 329–339.
- OpenAI. 2024. Data Processing Addendum. OpenAI. https://openai.com/policies/data-processing-addendum/ (Accessed on 05/09/2024).
- OpenAI. 2024. Plugin Terms. OpenAI. https://openai.com/policies/plugin-terms/ (Accessed on 05/09/2024).
- Google PAIR. 2019. People + AI Guidebook. Google Research. https://design.google/aiguidebook
- Florian Schaub, Rebecca Balebako, Adam L Durity, and Lorrie Faith Cranor. 2015. A design space for effective privacy notices. In Eleventh symposium on usable privacy and security (SOUPS 2015). USENIX Association, Ottawa, Canada, 1–17.
- Warda Usman, Jackie Hu, McKynlee Wilson, and Daniel Zappala. 2023. Distrust of big tech and a desire for privacy: Understanding the motivations of people who have voluntarily adopted secure email. In Nineteenth Symposium on Usable Privacy and Security (SOUPS 2023). USENIX, Anaheim, California, USA, 473–490.
- Heng Xu, Sumeet Gupta, Mary Beth Rosson, and John M. Carroll. 2012. Measuring Mobile Users’ Concerns for Information Privacy. In Proceedings of the 33rd International Conference on Information Systems (ICIS). Association for Information Systems, Orlando, FL, USA, 16 pages. https://aisel.aisnet.org/icis2012/proceedings/ISSecurity/10
Cited by
Understanding Users' Privacy Perceptions Towards LLM's RAG-based Memory
Zhang, S., Ma, R., Ma, Y., Li, S., Xu, Y., Yi, X., & Li, H. (2025) · Proceedings of the 2025 Workshop on Human-Centered AI Privacy and Security · doi:10.1145/3733816.3760750 · Cited by 6
“Users often rely on inaccurate mental models of LLM data flows and are vulnerable to dark patterns, which impedes their ability to grasp real privacy risks [22, 47].”
Self-citationTowards Aligning Personalized AI Agents with Users' Privacy Preference
Zhang, S., Ma, Y., Chen, J., Li, S., Yi, X., & Li, H. (2025) · doi:10.1145/3733816.3760752 · Cited by 19
“Furthermore, pioneering studies have explored methods for reminding users about the risks of geolocation inference from their social media data [26].”
Characterizing Unintended Consequences of GUI Agents For Web Browsing
Zhang, S., Chen, J., Gao, Z., Gao, J., Yi, X., & Li, H. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790696 · Cited by 19
“Unlike the episodic interaction of chat-bots [137] or simplistic agents [66], GUI agents require persistent login states to function across applications.”
Unsafe by Design? A First Look at Security and Privacy Risks in OpenAI’s Custom GPT Ecosystem
Ogundoyin, S. O., Ikram, M., Asghar, H. J., Zhao, B. Z. H., & Kâafar, M. A. (2025) · Workshop on Privacy in the Electronic Society · doi:10.1145/3733802.3764054 · Cited by 3
“Sub-mitting detailed reviews based on personal experience can surface hidden vulnerabilities and enhance collective safety [27] On the development side, creators must follow secure design principles across the GPT lifecycle, such as embedding protective prompt structures, minimizing permissions and…”
The Privacy Paradox of LLMs: User Perceptions and the Reality of PII Leakage
Cheng, S., Xu, H., Meng, S., Hao, S., Yue, C., & Li, Z. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791809 · Cited by 2
“Although existing user studies [5, 67, 72, 73, 75, 108, 122] have investigated LLM privacy and users’ understanding of PII, none have specifically addressed the critical issue of PII leakage in LLMs or investigated user perspectives concerning it.”
Metadata
- Authors
- Ma, R., Maidhof, C., Carrillo, J. C., Lindqvist, J., & Such, J.
- Affiliations
- Aalto University (FI)
- Universitat Politècnica de València (ES)
- King's College London, London, United Kingdom and VRAIN, Universitat Politècnica de València, Valencia, Spain
- Venue
- Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems
- Publisher
- ACM
- Published
- 2025-04-25
- Pages
- 1-18
- OpenAlex
- W4409883976
- Semantic Scholar
- 02d809e078953b5ef13a4f91b275ff1486fa6d17
- Cited by
- 8
- References
- 69
- Citations by year
- 2026: 5 · 2025: 3
- FWCI
- 26.87
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- CHI Conference on Human Factors in Computing Systems (publishedVersion, cc-by)
- Aaltodoc (Aalto University) (acceptedVersion, other-oa)
- Funders
- INCIBE (C063/23)
- Research Council of Finland (345991,345992)
- Topics
- Social Sciences › Sociology and Political Science › Privacy, Security, and Data Protection
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Keywords
- custom GPTs, perceived privacy, data flow, data handling, privacy regulation, proprietary knowledge, self-censorship, data transparency, platform regulation, large language models
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Privacy Perceptions of Custom GPTs by Users and Creators. https://app.cochat.ai/papers/10.1145/3706598.3713540
- The paper
- Ma, R., Maidhof, C., Carrillo, J. C., Lindqvist, J., & Such, J. (2025). Privacy Perceptions of Custom GPTs by Users and Creators. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3706598.3713540
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.

