- Rating
- 1
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 1
- Soundness
- 1
Read review
Paper summary
This paper presents an end to end reinforcement learning based safety alignment framework for tool using LLM agents. The authors identify two major threat vectors: user originated threats and tool originated threats. They propose a three way taxonomy (benign / malicious / sensitive) that applies both to user prompts and tool outputs,。Based the taxonomy,the paper apply an execute–refuse–verify policy to guide agent behavior. Training is conducted inside a custom sandbox environment with carefully designed reward functions.
Critical strengths
- The paper addresses an important and practically relevant problem in the safety alignment of tool-using language agents.
- Considering both user-side and tool-side threats provides a useful perspective on agent safety.
- The evaluation includes both security and utility benchmarks, which is important for assessing the practical trade-offs of safety alignment.
Critical weaknesses
- The paper claims establishing a unified tri-modal taxonomy as one of its main contributions. However, prior works (such as AgentAlign,Agent-SafetyBench) have already defined benign and harmful categories. Furthermore, the paper does not empirically justify the role of the newly added sensitive prompt category in experiments.
- There are multiple obvious writing errors. Rewards for Bu(214), Rewards for Bt(248), and including the corresponding equations that follow.
- The main baseline is only “no defense” and “prompt guarded”. The paper lacks comparison against recent agent safety alignment methods such as AgentAlign etc.
- Experiment results in table 2 and 3, the proposed method achieves strong performance on its self-constructed dataset, while the performance on existing public datasets is modest; in some cases it is even inferior to Prompt Guarded. The paper lacks corresponding analysis and does not release its constructed dataset.
- The paper states: “For B_U, we leverage datasets released by ReCall (Chen et al., 2025b). The remaining datasets are generated using Deepseek.” However, it does not provide detailed information regarding dataset generation.
Questions for authors
None