Refuse the Harm, Complete the Task via Graph-of-Rules Planning for LLM-Enabled Robots

Key points

• Large Language Models (LLMs) are increasingly used in embodied AI for robotic manipulation and planning, but their deployment in physical systems exposes vulnerabilities such as jailbreak attacks that induce unsafe behavior.1
• Existing inference-time defenses are mainly evaluated on short instructions containing a single hazard, whereas realistic tasks may interleave safe and unsafe sub-tasks, conceal hazardous steps, or combine individually safe actions into unsafe outcomes.1
• We introduce safe task completion, a task that requires an LLM-enabled robot to refuse the harmful part of an instruction while completing its legitimate part.1
• We evaluate RuleGraph on pure-attack and compound instructions across two benchmarks in two environments using Attack Success Rate (ASR) and Task Success Rate (TSR).1

Sources (1)

• [1]Refuse the Harm, Complete the Task via Graph-of-Rules Planning for LLM-Enabled Robots
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 17, 10:13 AM

  To address this problem, we propose RuleGraph, a multi-agent Graph-of-Rules planning framework that distributes safety enforcement across multiple reasoning and verification stages.

  Large Language Models (LLMs) are increasingly used in embodied AI for robotic manipulation and planning, but their deployment in physical systems exposes vulnerabilities such as jailbreak attacks that induce unsafe behavior.

Extractive summary: sentences quoted from the sources.