titleAgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security
authorsDongrui Liu, Yu Li, Zhonghao Yang, Peng Wang, Guanxu Chen, Yuejin Xie, Qinghua Mao, Wanying Qu, Yanxu Zhu, Tianyi Zhou, Leitao Yuan, Zhijie Zheng, Qihao Lin, Yimin Wang, Haoyu Luo, Shuai Shao, Chen Qian, Qingyu Liu, Ling Tang, Ruiyang Qin, Qihan Ren, Junxiao Yang, Kun Wang, Zhiheng Xi, Linfeng Zhang, Ranjie Duan, Bo Zhang, Wenjie Wang, Wen Shen, Qiaosheng Zhang, Yan Teng, Chaochao Lu, Rui Mei, Man Li, Jialing Tao, Xi Lin, Tianhang Zheng, Yong Liu, Quanshi Zhang, Lei Zhu, Xingjun Ma, Junhua Liu, Hui Xue, Xiaoxiang Zuo, Xiangnan He, Chao Shen, Xianglong Liu, Minlie Huang, Jing Shao, Xia Hu
date2026-05-28
venuearXiv (preprint, not peer reviewed)
idarXiv:2605.29801
affiliationshanghai ailab
upvotes144 (Hugging Face Daily Papers, 2026-11-01)
linksabs pdf hf
abstractModern 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.

cite

@misc{liu2026agentdog,
  title={AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security},
  author={Dongrui Liu and Yu Li and Zhonghao Yang and Peng Wang and Guanxu Chen and Yuejin Xie and Qinghua Mao and Wanying Qu and Yanxu Zhu and Tianyi Zhou and Leitao Yuan and Zhijie Zheng and Qihao Lin and Yimin Wang and Haoyu Luo and Shuai Shao and Chen Qian and Qingyu Liu and Ling Tang and Ruiyang Qin and Qihan Ren and Junxiao Yang and Kun Wang and Zhiheng Xi and Linfeng Zhang and Ranjie Duan and Bo Zhang and Wenjie Wang and Wen Shen and Qiaosheng Zhang and Yan Teng and Chaochao Lu and Rui Mei and Man Li and Jialing Tao and Xi Lin and Tianhang Zheng and Yong Liu and Quanshi Zhang and Lei Zhu and Xingjun Ma and Junhua Liu and Hui Xue and Xiaoxiang Zuo and Xiangnan He and Chao Shen and Xianglong Liu and Minlie Huang and Jing Shao and Xia Hu},
  year={2026},
  eprint={2605.29801},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2605.29801}
}