Zheng Lin

The paper introduces an embedding disruption method to re-activate and strengthen built-in safeguards within LLMs, effectively detecting and defending against sophisticated jailbreak attacks.

The paper introduces Disrupt-and-Rectify Smoothing (DR-Smoothing), a novel two-stage defense mechanism that significantly improves LLM security against jailbreaking attacks by restoring disrupted inputs to a safe, in-distribution form.

MemMark introduces a state-evolution attribution watermark that embeds owner-controlled signals into latent memory-write decisions, enabling robust provenance tracking for agent memory even when all traditional logs and metadata are lost.

This paper systematically analyzes how different architectural components of Large Vision-Language Models (LVLMs) contribute to hallucination robustness, finding that joint enhancement of visual fidelity and semantic alignment is most effective.