# OptiLeak: Efficient Prompt Reconstruction via Reinforcement Learning in Multi-tenant LLM Services

Source: [OptiLeak: Efficient Prompt Reconstruction via Reinforcement Learning in Multi-tenant LLM Services](<https://www.sota2.com/research/paper/699e81e66017c4460ffacd1b>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:00:04.574Z

Authors: Longxiang Wang, Xiang Zheng, Xuhao Zhang, Yao Zhang, Ye Wu, Cong Wang

Published: 2026-02-24T06:35:22.000Z

## Abstract

Multi-tenant LLM serving frameworks widely adopt shared Key-Value caches to enhance efficiency. However, this creates side-channel vulnerabilities enabling prompt leakage attacks. Prior studies identified these attack surfaces yet focused on expanding attack vectors rather than optimizing attack performance, reporting impractically high attack costs that underestimate the true privacy risk. We propose OptiLeak, a reinforcement learning-enhanced framework that maximizes prompt reconstruction efficiency through two-stage fine-tuning. Our key insight is that domain-specific \`\`hard tokens'' -- terms difficult to predict yet carrying sensitive information -- can be automatically identified via likelihood ranking and used to construct preference pairs for Direct Preference Optimization, eliminating manual annotation. This enables effective preference alignment while avoiding the overfitting issues of extended supervised fine-tuning. Evaluated on three benchmarks spanning medical and financial domains, OptiLeak achieves up to $12.48\\times$ reduction in average requests per token compared to baseline approaches, with consistent improvements across model scales from 3B to 14B parameters. Our findings demonstrate that cache-based prompt leakage poses a more severe threat than previously reported, underscoring the need for robust cache isolation in production deployments.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Prompt Leakage Attack on PubMedQA](<https://www.sota2.com/research/sota/prompt-leakage-attack-on-pubmedqa.md>): Prompt Leakage Attack; PubMedQA; ASR (500): 14.

- [Prompt Leakage Attack on MedQA](<https://www.sota2.com/research/sota/prompt-leakage-attack-on-medqa.md>): Prompt Leakage Attack; MedQA; ASR (500): 30.7.

- [Prompt Leakage Attack on FinanceBench](<https://www.sota2.com/research/sota/prompt-leakage-attack-on-financebench.md>): Prompt Leakage Attack; FinanceBench; ASR500: 88.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.20595>)

[PDF](<https://arxiv.org/pdf/2602.20595.pdf>)