Research

Paper

AI LLM March 17, 2026

Rotated Robustness: A Training-Free Defense against Bit-Flip Attacks on Large Language Models

Authors

Deng Liu, Song Chen

Abstract

Hardware faults, specifically bit-flips in quantized weights, pose a severe reliability threat to Large Language Models (LLMs), often triggering catastrophic model collapses. We demonstrate that this vulnerability fundamentally stems from the spatial alignment between sensitive weight bits and extreme activation outliers, which causes a single hardware fault to be massively amplified. To address this, we propose Rotated Robustness (RoR), a training-free defense utilizing orthogonal Householder transformations. By applying an orthogonal rotation to the activation space, RoR geometrically smooths extreme outliers across all feature dimensions. This mechanism effectively breaks the alignment between outliers and vulnerable weights, mathematically guaranteeing original model accuracy. Extensive empirical evaluations across Llama-2/3, OPT, and Qwen families demonstrate the superior reliability of our approach. Under random bit-flip attacks, RoR reduces the stochastic collapse rate from 3.15\% to 0.00\% on Qwen2.5-7B. Furthermore, under severe targeted attacks with 50 Progressive Bit Search flips, RoR sustains robust reasoning on Llama-2-7B, maintaining a 43.9\% MMLU accuracy that nearly matches its 45.2\% unattacked accuracy, while competing defenses collapse to random guessing. Most notably, against the Single-Point Fault Attack (SPFA) -- the most aggressive targeted threat -- RoR exponentially inflates the attack complexity from a few bits to over 17,000 precise bit-flips. With a negligible storage overhead of 0.31\% and a minimal inference latency increase of 9.1\% on Llama-2-7B, RoR achieves true lossless robustness, providing a practical and highly reliable defense for LLM deployment.

Metadata

arXiv ID: 2603.16382
Provider: ARXIV
Primary Category: cs.CR
Published: 2026-03-17
Fetched: 2026-03-18 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.16382v1</id>\n    <title>Rotated Robustness: A Training-Free Defense against Bit-Flip Attacks on Large Language Models</title>\n    <updated>2026-03-17T11:11:17Z</updated>\n    <link href='https://arxiv.org/abs/2603.16382v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.16382v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Hardware faults, specifically bit-flips in quantized weights, pose a severe reliability threat to Large Language Models (LLMs), often triggering catastrophic model collapses. We demonstrate that this vulnerability fundamentally stems from the spatial alignment between sensitive weight bits and extreme activation outliers, which causes a single hardware fault to be massively amplified. To address this, we propose Rotated Robustness (RoR), a training-free defense utilizing orthogonal Householder transformations. By applying an orthogonal rotation to the activation space, RoR geometrically smooths extreme outliers across all feature dimensions. This mechanism effectively breaks the alignment between outliers and vulnerable weights, mathematically guaranteeing original model accuracy. Extensive empirical evaluations across Llama-2/3, OPT, and Qwen families demonstrate the superior reliability of our approach. Under random bit-flip attacks, RoR reduces the stochastic collapse rate from 3.15\\% to 0.00\\% on Qwen2.5-7B. Furthermore, under severe targeted attacks with 50 Progressive Bit Search flips, RoR sustains robust reasoning on Llama-2-7B, maintaining a 43.9\\% MMLU accuracy that nearly matches its 45.2\\% unattacked accuracy, while competing defenses collapse to random guessing. Most notably, against the Single-Point Fault Attack (SPFA) -- the most aggressive targeted threat -- RoR exponentially inflates the attack complexity from a few bits to over 17,000 precise bit-flips. With a negligible storage overhead of 0.31\\% and a minimal inference latency increase of 9.1\\% on Llama-2-7B, RoR achieves true lossless robustness, providing a practical and highly reliable defense for LLM deployment.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CR'/>\n    <published>2026-03-17T11:11:17Z</published>\n    <arxiv:comment>13 pages, 8 figures. Preprint. Under review at IEEE Transactions on Information Forensics and Security</arxiv:comment>\n    <arxiv:primary_category term='cs.CR'/>\n    <author>\n      <name>Deng Liu</name>\n    </author>\n    <author>\n      <name>Song Chen</name>\n    </author>\n  </entry>"
}