Preprint · 2026 · Open access
Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory
arXiv (Cornell University) · Published 20 October 2026 · DOI 10.48550/arxiv.2610.28315
In brief
“Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory” is a 2026 preprint by Masahiro Kaneko and 2 co-authors, published in arXiv (Cornell University). OpenAlex counts 0 citations as of 19 November 2026. It is open access under a CC BY licence: the full text is free to read at arXiv (Cornell University).
- Citations
- 0
- References
- 39
- Authors
- 3
Abstract
Large language model (LLM) agents increasingly use persistent memory to personalize responses, and personal attributes in this memory pose a privacy risk if exposed. We ask how consulting memory during dialogue changes what users go on to say, and thus what the memory comes to reveal. In multi-session simulations, a simulated user with 24 personal attributes talks with a support agent that does or does not consult memory, and in both settings attributes are inferred from the full accumulated record. Across four support models, memory raises response quality but also empirical leakage, the share of attribute values correctly inferred from the record, by a margin that widens over sessions. The cause is a loop in ordinary, non-adversarial assistance. Memory-conditioned responses carry earlier values forward, often as presuppositions, and users affirm them or leave them uncorrected, so that cues and the agent's own guesses become recorded values. The loop reinforces itself, and removing the agent's mentions of an attribute lowers its later disclosure about twice as much as single-turn effects predict. Users cannot leave it on their own, since correcting a value discloses the attribute, declining to discuss it shifts later predictions toward sensitive values, and incorrect values left in place become entrenched. The loop is best weakened at the agent's response. Instructing the agent not to state or presuppose attributes brings leakage to about the no-memory level while keeping most of the quality gains, whereas filtering or anonymizing what users say or what is stored gives up much of them.
Cite this paper
Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.
APA 7
Kaneko, M., Arai, H., & Baldwin, T. (2026). Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2610.28315
MLA 9
Kaneko, Masahiro, et al. “Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory.” arXiv (Cornell University), 2026. https://doi.org/10.48550/arxiv.2610.28315.
Chicago
Kaneko, Masahiro, Hiromi Arai, and Timothy Baldwin. 2026. “Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2610.28315.
Harvard
Kaneko, M., Arai, H. and Baldwin, T. (2026) ‘Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory’, arXiv (Cornell University). doi:10.48550/arxiv.2610.28315.
IEEE
M. Kaneko, H. Arai, and T. Baldwin, “Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory,” arXiv (Cornell University), 2026, doi: 10.48550/arxiv.2610.28315.
BibTeX
@misc{kaneko2026disclosure,
title = {{Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory}},
author = {Kaneko, Masahiro and Arai, Hiromi and Baldwin, Timothy},
howpublished = {arXiv (Cornell University)},
year = {2026},
doi = {10.48550/arxiv.2610.28315},
url = {https://doi.org/10.48550/arxiv.2610.28315}
}
Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)
Topic and keywords
Social Sciences › Social Sciences › Sociology and Political Science › Privacy, Security, and Data Protection
- LLM agents
- persistent memory
- privacy leakage
- self-reinforcing loops
- presupposition
- data anonymization
- response quality
Work with this paper
Most-cited papers in Social Sciences
- Using thematic analysis in psychologyVirginia Braun, Victoria Clarke · Qualitative Research in Psychology · 2006 · 240,222 citations
- Case Study Research: Design and MethodsRobert K. Yin · 1984 · 81,951 citations
- Three Approaches to Qualitative Content AnalysisHsiu-Fang Hsieh, Sarah Elizabeth Shannon · Qualitative Health Research · 2005 · 49,623 citations
- Qualitative Data Analysis: An Expanded SourcebookMatthew B. Miles, A Michael Huberman · Journal of Environmental Psychology · 1994 · 48,311 citations
- Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groupsAllison Tong, Peter Sainsbury, Jonathan C. Craig · International Journal for Quality in Health Care · 2007 · 45,353 citations
Questions about this paper
How do I cite this paper in APA style?
Kaneko, M., Arai, H., & Baldwin, T. (2026). Disclosure Begets Disclosure: A Self-Reinforcing Privacy Loop in LLM Agents with Persistent Memory. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2610.28315
How many times has this paper been cited?
0 times, according to OpenAlex on 19 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.
Is this paper free to read?
Yes. It is open access at arXiv (Cornell University): sign in and open the PDF from this page.
Who wrote this paper?
Masahiro Kaneko, Hiromi Arai, Timothy Baldwin.
What is the DOI of this paper?
10.48550/arxiv.2610.28315, permanently at https://doi.org/10.48550/arxiv.2610.28315. Sign in and open it from this page.
Bibliographic data from OpenAlex (CC0), last checked 19 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.