Research map / Large language models / Instruction and fine-tuning
Durable Quantization Conditioned Misalignment Attack on Large Language Models
Peiran Dong, Haowei Li, Song Guo
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Introduces Q-Misalign, an attack that hides safety misalignment in a full-precision LLM so it only surfaces after quantization, with the misalignment designed to persist through subsequent fine-tuning.
Problem
Quantized LLMs deployed on edge devices may expose new safety vulnerabilities; the paper targets misalignment that stays dormant in the full-precision model and activates after quantization.
Approach
Safety misalignment is embedded so that it remains inactive in the full-precision model and is triggered only by the quantization step, framed as a quantization-conditioned attack that is durable under further fine-tuning.
Closest papers by meaning
- Exploiting LLM Quantization (NeurIPS 2024 · similarity 0.91)
- Assessing Safety Risks and Quantization-aware Safety Patching for Quantized Large Language Models (ICML 2025 · similarity 0.90)
- Safety Alignment Should be Made More Than Just a Few Tokens Deep (ICLR 2025 · similarity 0.89)
- Safety Layers in Aligned Large Language Models: The Key to LLM Security (ICLR 2025 · similarity 0.89)
- Fail-Closed Alignment for Large Language Models (NeurIPS 2026 · similarity 0.88)
- GuardAlign: Robust Safety Alignment in Multimodal Large Language Models (ICLR 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
