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

Hugging Face Daily Papers Papers
agent-safety alignment lightweight scalable ai-agents safety-framework open-source

Summary

This paper proposes AgentDoG 1.5, a lightweight and scalable alignment framework for AI agent safety, using taxonomy-guided training with minimal samples to achieve performance comparable to leading closed-source models.

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.
Original Article
View Cached Full Text

Cached at: 05/29/26, 02:59 AM

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

Source: https://huggingface.co/papers/2605.29801 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

A lightweight and scalable agent safety alignment framework is proposed to address emerging threats from advanced AI models, featuring taxonomy-guided training with minimal samples and efficient deployment in real-world scenarios.

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 scalableagent safety alignment framework. Specifically, we update theagent safety taxonomyto accommodate emergent risks from Codex and OpenClaw execution scenarios. We further build a taxonomy-guided data engine withinfluence-function purificationto train lightweightAgentDoG 1.5variants (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 onAgentDoG 1.5, we construct a highly efficientagentic safety SFTandRL training environment, which reduces deployment overhead inDocker-level environmentsby two orders of magnitude. Finally, we deployAgentDoG 1.5as a training-freeonline guardrailforreal-time safety moderation. Extensive experimental results indicate thatAgentDoG 1.5achieves state-of-the-art performance in diverse and complexinteractive agentic scenarios. All models and datasets are openly released.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2605\.29801

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2605.29801 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.29801 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.29801 in a Space README.md to link it from this page.

Collections including this paper1

Similar Articles

DOG-DPO:Dynamic Optimization in Geometry for Safety Alignment

arXiv cs.LG

DOG-DPO is a training-free data selection framework that treats preference pairs as structured geometric signals, decomposing multi-dataset preference geometry into anchor and residual subspaces to select diverse subsets for safety alignment. It achieves strong utility-robustness trade-offs using only 11% of preference pairs across six safety benchmarks.

Decoupled Alignment for Robust Plug-and-Play Adaptation

arXiv cs.CL

Introduces a training-free method for enhancing safety alignment of LLMs by using knowledge distillation and model fusion to prevent shadow alignment, improving defense success rate by 14.42% on harmful question datasets without compromising performance.