StepGuard: Learning Step-Level Guardrails with Scalable Supervision and Safety-Utility Balancing

StepGuard introduces a step-level guard model that audits and checks tool actions before execution, addressing security risks in LLM-based agents. It is trained using StepGen, an automatic engine that generates safe and unsafe trajectories, and employs Balance-GRPO to dynamically balance learning between safe and unsafe actions. Experiments show StepGuard achieves high accuracy comparable to GPT-5.4 and significantly reduces attack success rates while minimally impacting utility.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• llms
• agents
• safety