New 'State Backdoor' attack targets embodied AI models

Researchers have developed a new type of backdoor attack targeting Vision-Language-Action (VLA) models, which are crucial for embodied AI applications like robotics. Unlike previous methods that rely on visible visual triggers, this novel "State Backdoor" utilizes the initial state of a robot arm as the trigger. A Preference-guided Genetic Algorithm was employed to find minimal yet effective state-based triggers, achieving over 90% attack success without degrading performance on normal tasks. AI

IMPACT Reveals a new vulnerability in embodied AI, potentially requiring new security measures for robotic systems.

RANK_REASON The cluster contains a research paper detailing a novel attack method on AI models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Preference-guided Genetic Algorithm
• State Backdoor
• Vision-Language-Action models
• Yansong Lin

• safety
• paper

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