New attack hijacks MLLMs with single perturbation · arXiv research

Researchers have developed a novel attack method called Semantic-Aware Hijacking that can compromise Multimodal Large Language Models (MLLMs) with a single adversarial perturbation. This technique, termed Semantic-Aware Universal Perturbation (SAUP), functions as a semantic router, directing inputs to attacker-defined targets. Experiments on models like Qwen demonstrated a 66% success rate in hijacking five distinct targets with a single perturbation. AI

IMPACT This research highlights a significant vulnerability in MLLMs, potentially impacting their deployment in safety-critical applications like autonomous driving and robotics.

RANK_REASON Research paper detailing a novel attack on MLLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Changyue Li
• MLLMs
• Qwen
• Semantic-Aware Hijacking
• Semantic-Aware Universal Perturbation (SAUP)
• Semantic Router

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →