5 papers
Defense Against LLM Backdoors using Critical Neuron Isolation Pruning
Yuxi Li, Zhibo Zhang, Kailong Wang +3
Large language models (LLMs) are vulnerable to backdoor attacks, where hidden triggers induce malicious outputs. Existing defenses generally fall into inference-time detection or t…
Understanding Safety-Sensitive Expert Behavior in Mixture-of-Experts LLMs
Zhibo Zhang, Yuxi Li, Zhen Ouyang +2
Mixture-of-Experts (MoE) LLMs rely on sparse, router-driven expert activation, yet how safety alignment interacts with routed expert specialization remains underexplored. A common…
Uncovering Logit Suppression Vulnerabilities in LLM Safety Alignment
Yuxi Li, Yi Liu, Yuekang Li +4
Large language models (LLMs) have revolutionized various applications, making robust safety alignment essential to prevent harmful outputs. Current safety alignment techniques, how…
Embedding Poisoning: Bypassing Safety Alignment via Embedding Semantic Shift
Shuai Yuan, Zhibo Zhang, Yuxi Li +2
The widespread distribution of Large Language Models (LLMs) through public platforms like Hugging Face introduces significant security challenges. While these platforms perform bas…
Circumventing Safety Alignment in Large Language Models Through Embedding Space Toxicity Attenuation
Zhibo Zhang, Yuxi Li, Kailong Wang +3
Large Language Models (LLMs) have achieved remarkable success across domains such as healthcare, education, and cybersecurity. However, this openness also introduces significant se…