Preprint · 2026 · Open access
SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs
arXiv (Cornell University) · Published 30 September 2026 · DOI 10.48550/arxiv.2609.38830
In brief
“SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs” is a 2026 preprint by Fahao Chen and 4 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
- 44
- Authors
- 5
Abstract
Sparse attention is widely used to accelerate long-context inference in modern large language models (LLMs), but its input-dependent execution behavior introduces previously unexplored privacy risks. We identify a new GPU micro-architectural side channel, termed Sparsity-Induced Memory Access (SIMA), which arises from secret-dependent key-value cache access patterns induced by sparse attention. Based on this observation, we present SparLeak, a phase-aware side-channel attack that extracts SIMA traces during LLM inference and enables two practical privacy extractions: query attribute inference from prefill-phase traces and autoregressive response reconstruction from decoding-phase traces. By reconstructing approximate token-level sparsity profiles from page-level observations and applying profiling-based learning, SparLeak accurately recovers sensitive information, including user-query attributes and private LLM response content. Extensive evaluation across three LLM architectures, three sparse attention mechanisms, and three privacy-sensitive datasets shows that SparLeak achieves average attack success rates of 90.9% for attribute inference and 87.3% for response reconstruction under real-world LLM serving settings, highlighting the significance to account for SIMA leakage when deploying sparse-attention-based LLM systems. We provide anonymized SIMA traces, trained attack models, evaluation scripts, and documentation as artifacts at https://anonymous.4open.science/r/Janus_artifacts/.
Cite this paper
Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.
APA 7
Chen, F., Du, L., Zhou, J., Li, P., & Su, Z. (2026). SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2609.38830
MLA 9
Chen, Fahao, et al. “SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs.” arXiv (Cornell University), 2026. https://doi.org/10.48550/arxiv.2609.38830.
Chicago
Chen, Fahao, Linkang Du, Jinhao Zhou, Peng Li, and Zhou Su. 2026. “SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2609.38830.
Harvard
Chen, F. et al. (2026) ‘SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs’, arXiv (Cornell University). doi:10.48550/arxiv.2609.38830.
IEEE
F. Chen, L. Du, J. Zhou, P. Li, and Z. Su, “SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs,” arXiv (Cornell University), 2026, doi: 10.48550/arxiv.2609.38830.
BibTeX
@misc{chen2026sparleak,
title = {{SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs}},
author = {Chen, Fahao and Du, Linkang and Zhou, Jinhao and Li, Peng and Su, Zhou},
howpublished = {arXiv (Cornell University)},
year = {2026},
doi = {10.48550/arxiv.2609.38830},
url = {https://doi.org/10.48550/arxiv.2609.38830}
}
Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)
Topic and keywords
Physical Sciences › Computer Science › Hardware and Architecture › Security and Verification in Computing
- sparse attention
- large language models
- privacy leakage
- large language model inference
- response reconstruction
- long-context inference
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?
Chen, F., Du, L., Zhou, J., Li, P., & Su, Z. (2026). SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2609.38830
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?
Fahao Chen, Linkang Du, Jinhao Zhou, Peng Li, Zhou Su.
What is the DOI of this paper?
10.48550/arxiv.2609.38830, permanently at https://doi.org/10.48550/arxiv.2609.38830. 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.