arXiv · cs.AI · Anes Abdennebi, Nadjia Kara, Laaziz Lahlou, Hakima Ould-Slimane2026-11-02 13:00 · 9 小时前模型评分56

LanG 2.0 提出用小型本地模型做大规模安全运营的智能体系统

LanG 2.0: Energy-Efficient Agentic AI for Semi-Automated Security Operations at Scale

导读

LanG 2.0 是一个面向安全运营中心（SOC，Security Operations Center）的智能体系统，负责检测威胁、关联证据并编写规则，研究小型本地模型能否在结构性防护而非模型准确率提供保障的前提下，低成本且安全地替代前沿模型。

分类置信度

中 · 待人工复核

来源级别：媒体与个人

应用领域：政务与公共事务

事件类型：研究发布

证据摘录

“We present LanG 2.0, an agentic system that detects threats, correlates evidence and writes rules, investigating whether small local models can replace frontier models affordably and safely when structural safeguards, not model accuracy, provide the guarantees.”
“Finetuned models statistically match it on alert classification and rule writing for 5.6 to 59 times less energy.”

来源：arXiv · cs.AI · arxiv.org