Preprint · 2026 · Open access
Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing
arXiv (Cornell University) · Published 26 April 2026 · DOI 10.48550/arxiv.2604.23711
In brief
“Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing” is a 2026 preprint by Yu Cui and 8 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
- 32
- Authors
- 9
Abstract
With the wide adoption of personal AI assistants such as OpenClaw, privacy leakage in user interaction contexts with large language model (LLM) agents has become a critical issue. Existing privacy attacks against LLMs primarily target training data, while research on inference-time contextual privacy risks in LLM agent memory remains limited. Moreover, prior methods often incur high attack costs, requiring multiple queries or relying on white-box assumptions, which limits their practicality in real-world deployments. To address these issues, we propose a training-free privacy extraction attack targeting LLM agent memory, which we name \textsc{Spore}. \textsc{Spore} is compatible with both black-box and gray-box settings. In the black-box setting, \textsc{Spore} can efficiently extract a small candidate set via a single query to recover the original private information. In the gray-box setting, \textsc{Spore} allows the attacker to leverage multi-ranked tokens for more accurate and faster privacy extraction. We provide an information-theoretic analysis of \textsc{Spore} and show that it achieves high query efficiency with substantial per query information leakage. Experiments on multiple frontier LLMs show that \textsc{Spore} outperforms attack success rate over existing state-of-the-art (SOTA) schemes. It also maintains low attack cost and remains stable across different model parameter settings. We further evaluate the robustness of \textsc{Spore} against existing defense mechanisms. Our results show that \textsc{Spore} consistently bypasses both detection and strong safety alignment, demonstrating resilient performance in diverse defensive settings and real-world safety threats.
Cite this paper
Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.
APA 7
Cui, Y., Yue, R., Fu, H., Pan, S., Zhuoyu, S., Huang, B., Zhang, H., Zuo, C., & Wang, L. (2026). Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2604.23711
MLA 9
Cui, Yu, et al. “Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing.” arXiv (Cornell University), 2026. https://doi.org/10.48550/arxiv.2604.23711.
Chicago
Cui, Yu, Ruiqing Yue, Hang Fu, Sicheng Pan, Sun Zhuoyu, Baohan Huang, Haibin Zhang, Cong Zuo, and Licheng Wang. 2026. “Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2604.23711.
Harvard
Cui, Y. et al. (2026) ‘Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing’, arXiv (Cornell University). doi:10.48550/arxiv.2604.23711.
IEEE
Y. Cui et al., “Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing,” arXiv (Cornell University), 2026, doi: 10.48550/arxiv.2604.23711.
BibTeX
@misc{cui2026spore,
title = {{Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing}},
author = {Cui, Yu and Yue, Ruiqing and Fu, Hang and Pan, Sicheng and Zhuoyu, Sun and Huang, Baohan and Zhang, Haibin and Zuo, Cong and Wang, Licheng},
howpublished = {arXiv (Cornell University)},
year = {2026},
doi = {10.48550/arxiv.2604.23711},
url = {https://doi.org/10.48550/arxiv.2604.23711}
}
Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)
Topic and keywords
Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- LLM agents
- black-box attacks
- gray-box attacks
- information-theoretic analysis
- attack cost
- defense evasion
- safety alignment
- personal AI assistants
Work with this paper
Most-cited papers in Computer Science
- ImageNet classification with deep convolutional neural networksAlex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton · Communications of the ACM · 2017 · 114,726 citations
- Long Short-Term MemorySepp Hochreiter, Jürgen Schmidhuber · Neural Computation · 1997 · 109,097 citations
- A Mathematical Theory of CommunicationClaude E. Shannon · Bell System Technical Journal · 1948 · 91,736 citations
- Deep learningYann LeCun, Yoshua Bengio, Geoffrey E. Hinton · Nature · 2015 · 90,226 citations
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal NetworksShaoqing Ren, Kaiming He, Ross Girshick and 1 more · IEEE Transactions on Pattern Analysis and Machine Intelligence · 2016 · 74,143 citations
Questions about this paper
How do I cite this paper in APA style?
Cui, Y., Yue, R., Fu, H., Pan, S., Zhuoyu, S., Huang, B., Zhang, H., Zuo, C., & Wang, L. (2026). Spore: Efficient and Training-Free Privacy Extraction Attack on LLMs via Inference-Time Hybrid Probing. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2604.23711
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?
Yu Cui, Ruiqing Yue, Hang Fu, Sicheng Pan, Sun Zhuoyu, Baohan Huang, Haibin Zhang, Cong Zuo, Licheng Wang.
What is the DOI of this paper?
10.48550/arxiv.2604.23711, permanently at https://doi.org/10.48550/arxiv.2604.23711. 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.