SAFETY SENTRY: Context-Aware Human Intervention via EXECUTE-ASK-REFUSE Routing
Quick Answer
Safety Sentry introduces a three-way routing decision model (EXECUTE, ASK, REFUSE) for LLM agents, enhancing safety by contextualizing actions.
Quick Take
Safety Sentry introduces a three-way routing decision model (EXECUTE, ASK, REFUSE) for LLM agents, enhancing safety by contextualizing actions. It outperforms existing models in accuracy and safety recall while allowing flexible deployment across varying risk tolerances without retraining.
Key Points
- Safety Sentry uses a lightweight guard model for real-time decision-making.
- It distinguishes between harmful actions and context-appropriate actions.
- The model achieves superior accuracy and safety recall compared to existing baselines.
- A single decoding-time threshold allows for adaptable risk management.
- Routine interruptions are minimized, preserving user autonomy.
Paper Resources
📖 Reader Mode
~2 min readAbstract:LLM agents act on real-world environments through tool calls, and a single misjudged action can cause irreversible harm. The standard safeguard is a guard model that labels each proposed action as safe or unsafe, but this binary view conflates two distinct decisions: whether the action is harmful in itself, and whether it is appropriate given the user's context. It also operates at the granularity of action categories rather than individual instances, producing routine interruptions that erode autonomy and train users to wave through the most consequential alerts. We reframe the problem as a per-instance three-way routing decision over {EXECUTE, ASK, REFUSE} and instantiate it with Safety Sentry, a lightweight guard model whose inference reduces to a single decoding call. A single decoding-time threshold lets one fixed checkpoint be re-positioned across deployments of differing risk tolerance without retraining. Safety Sentry outperforms a broad set of open-weight and frontier closed-source baselines on overall accuracy and safety-related recall, while controlling both directional error rates simultaneously.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.13594 [cs.AI] |
| (or arXiv:2607.13594v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.13594 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Chujia Hu [view email]
[v1]
Wed, 15 Jul 2026 08:38:16 UTC (1,220 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →Automatic Ordinary Differential Equations Discovery For Biological Systems Using Large Language Model Powered Agentic System
The MEDA system utilizes large language models and symbolic regression to autonomously discover ordinary differential equations for biological systems, achieving strong structural recovery and biologically plausible models. It outperforms existing methods by integrating domain knowledge and mechanistic constraints, demonstrating effective retrieval and extrapolation capabilities.