Preprint · 2026 · Open access

HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents

Lingxiang Xu, Jiaoyun Yang, Min Hu, Ning An

arXiv (Cornell University) · Published 4 June 2026 · DOI 10.48550/arxiv.2606.06055

Open PDFView at the publisherCite this paper

In brief

“HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents” is a 2026 preprint by Lingxiang Xu and 3 co-authors, published in arXiv (Cornell University). OpenAlex counts 2 citations as of 19 November 2026. It is open access: the full text is free to read at arXiv (Cornell University).

Citations
2
Most cited in
2026 (2)
References
43
Authors
4

Abstract

Long-term memory helps conversational agents maintain continuity across sessions, while relevance and current-turn warrant remain distinct decisions. We study this boundary under a stated conservative policy in which sensitive history shapes a response when the current turn supplies a reason to use it. We introduce HUSH-Bench, a controlled benchmark of 2,400 benign prompts paired with histories containing one marked sensitive disclosure and matched no-memory references. HUSH-Bench measures unsolicited history integration with the Unsolicited History Integration Score (UIS; 0--100, higher is worse), records whether the marked disclosure reaches the generator, and includes paired prompts that differ only in whether the user asks the assistant to use earlier context. We evaluate four models under no-memory, full-context, and three retrieval-based memory settings. Memory access raises UIS from near zero to 8.9--26.6 for one model and 51.3--83.0 for the other three. Retrieval systems expose the marked disclosure in 23.0\%--30.3\% of cases, while related sensitive entries or summaries remain available and three models continue to show high UIS. Across four generators, an explicit invitation increases target-memory uptake scores by 27.0--41.3; measured helpfulness remains stable while mean over-scope rises. These results motivate treating memory storage, retrieval, warrant, and per-turn scope as separate design decisions.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Xu, L., Yang, J., Hu, M., & An, N. (2026). HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2606.06055

MLA 9

Xu, Lingxiang, et al. “HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents.” arXiv (Cornell University), 2026. https://doi.org/10.48550/arxiv.2606.06055.

Chicago

Xu, Lingxiang, Jiaoyun Yang, Min Hu, and Ning An. 2026. “HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2606.06055.

Harvard

Xu, L. et al. (2026) ‘HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents’, arXiv (Cornell University). doi:10.48550/arxiv.2606.06055.

IEEE

L. Xu, J. Yang, M. Hu, and N. An, “HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents,” arXiv (Cornell University), 2026, doi: 10.48550/arxiv.2606.06055.

BibTeX

@misc{xu2026hush,
  title     = {{HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents}},
  author    = {Xu, Lingxiang and Yang, Jiaoyun and Hu, Min and An, Ning},
  howpublished = {arXiv (Cornell University)},
  year      = {2026},
  doi       = {10.48550/arxiv.2606.06055},
  url       = {https://doi.org/10.48550/arxiv.2606.06055}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems

  • chatbots
  • long-term memory
  • warrants
  • benchmark

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Xu, L., Yang, J., Hu, M., & An, N. (2026). HUSH-Bench: Measuring Memory-Use Boundaries for Sensitive History in Conversational Agents. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2606.06055

How many times has this paper been cited?

2 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?

Lingxiang Xu, Jiaoyun Yang, Min Hu, Ning An.

What is the DOI of this paper?

10.48550/arxiv.2606.06055, permanently at https://doi.org/10.48550/arxiv.2606.06055. 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.