Saved in:
Bibliographic Details
Main Authors: Yan, Xiaobei, Li, Yiming, Wang, Hao, Qiu, Han, Zhang, Tianwei
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.16670
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914340425170944
author Yan, Xiaobei
Li, Yiming
Wang, Hao
Qiu, Han
Zhang, Tianwei
author_facet Yan, Xiaobei
Li, Yiming
Wang, Hao
Qiu, Han
Zhang, Tianwei
contents Large language models (LLMs) are widely deployed, but their substantial compute demands make them vulnerable to inference cost attacks that aim to deliberately maximize the output length. In this work, we investigate a distinct attack surface: maximizing inference cost by tampering with the model parameters instead of inputs. This approach leverages the established capability of Bit-Flip Attacks (BFAs) to persistently alter model behavior via minute weight perturbations, effectively decoupling the attack from specific input queries. To realize this, we propose BitHydra, a framework that addresses the unique optimization challenge of identifying the exact weight bits that maximize generation cost. We formulate the attack as a constrained Binary Integer Programming (BIP) problem designed to systematically suppress the end-of-sequence (i.e., <eos>) probability. To overcome the intractability of the discrete search space, we relax the problem into a continuous optimization task and solve it via the Alternating Direction Method of Multipliers (ADMM). We evaluate BitHydra across 10 LLMs (1.5B-16B). Our results demonstrate that the proposed optimization method efficiently achieves endless generation with as few as 1-4 bit flips on all testing models, verifying the effectiveness of the ADMM-based formulation against both standard models and potential defenses.
format Preprint
id arxiv_https___arxiv_org_abs_2505_16670
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle BitHydra: Towards Bit-flip Inference Cost Attack against Large Language Models
Yan, Xiaobei
Li, Yiming
Wang, Hao
Qiu, Han
Zhang, Tianwei
Cryptography and Security
Artificial Intelligence
Large language models (LLMs) are widely deployed, but their substantial compute demands make them vulnerable to inference cost attacks that aim to deliberately maximize the output length. In this work, we investigate a distinct attack surface: maximizing inference cost by tampering with the model parameters instead of inputs. This approach leverages the established capability of Bit-Flip Attacks (BFAs) to persistently alter model behavior via minute weight perturbations, effectively decoupling the attack from specific input queries. To realize this, we propose BitHydra, a framework that addresses the unique optimization challenge of identifying the exact weight bits that maximize generation cost. We formulate the attack as a constrained Binary Integer Programming (BIP) problem designed to systematically suppress the end-of-sequence (i.e., <eos>) probability. To overcome the intractability of the discrete search space, we relax the problem into a continuous optimization task and solve it via the Alternating Direction Method of Multipliers (ADMM). We evaluate BitHydra across 10 LLMs (1.5B-16B). Our results demonstrate that the proposed optimization method efficiently achieves endless generation with as few as 1-4 bit flips on all testing models, verifying the effectiveness of the ADMM-based formulation against both standard models and potential defenses.
title BitHydra: Towards Bit-flip Inference Cost Attack against Large Language Models
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2505.16670