The Year in AI Papers
The year/Independent research

Paper 2605.29801

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

Dongrui Liu, Yu Li, Zhonghao Yang, Peng Wang, Guanxu Chen, Yuejin Xie, Qinghua Mao, Wanying Qu, Yanxu Zhu, Tianyi Zhou, Leitao Yuan, Zhijie Zheng, et al.

Published
May 2026
Research lab
Independent
Citations
2
GitHub
Not linked

01 In brief

Summary

AgentDoG 1.5, developed by Shanghai Artificial Intelligence Laboratory, is a lightweight and scalable framework for AI agent safety and security.

It updates the three-dimensional safety taxonomy (risk source, failure mode, real-world harm) to cover new risks from Codex and OpenClaw execution scenarios, extending the ATBench benchmark family with ATBench-Codex and ATBench-Claw.

Using a taxonomy-guided data engine with influence-function purification, AgentDoG 1.5 variants (0.8B, 2B, 4B, 8B) are trained on only ~1k samples, achieving performance comparable to closed-source models like GPT-5.4.

The framework includes a highly efficient agentic safety SFT and RL training environment, reducing deployment overhead by two orders of magnitude compared to Docker-level environments, and supports over 10,000 concurrent environments on an 8-core machine.

AgentDoG 1.5 also serves as a training-free online guardrail for real-time safety moderation.

Extensive experiments show state-of-the-art performance on benchmarks like R-Judge and ATBench, with AgentDoG 1.5-4B achieving 92.2% accuracy on R-Judge and 72.4% on ATBench, and reducing unsafe final deliveries in OpenClaw agents.

All models and datasets are openly released.

02 From the paper

Abstract

Modern open-world agents such as OpenClaw exhibit powerful cross-environment execution capabilities yet introduce broad new safety risk sources. Meanwhile, advanced frontier AI models drastically lower attack barriers, rendering current agent alignment frameworks inadequate for real-world deployment. To tackle these emerging threats, we propose a lightweight and scalable agent safety alignment framework. Specifically, we update the agent safety taxonomy to accommodate emergent risks from Codex and OpenClaw execution scenarios. We further build a taxonomy-guided data engine with influence-function purification to train lightweight AgentDoG 1.5 variants (0.8B, 2B, 4B, and 8B parameters) using only around 1k samples, achieving comparable performance with leading closed-source models (e.g., GPT-5.4). Based on AgentDoG 1.5, we construct a highly efficient agentic safety SFT and RL training environment, which reduces deployment overhead in Docker-level environments by two orders of magnitude. Finally, we deploy AgentDoG 1.5 as a training-free online guardrail for real-time safety moderation. Extensive experimental results indicate that AgentDoG 1.5 achieves state-of-the-art performance in diverse and complex interactive agentic scenarios. All models and datasets are openly released.