风云卫AI赋能安全系列
清风卫AI自身安全系列
基础设施安全
数据安全
云计算安全
工业互联网安全
物联网安全
信息技术应用创新
全部产品
全部解决方案
绿盟大模型红队评估服务
NSFOCUS LLM Red Team Assessment Service
从攻击者视角检验大模型与智能体的业务安全风险
Adversarial assessment of LLMs and AI agents in business applications
绿盟大模型红队评估服务面向大模型及智能体接入业务后的安全风险。服务结合自动化测试与攻防专家人工测试,从攻击者视角检验模型、应用逻辑和底层基础设施的安全性。服务参考MITRE ATLAS等对抗攻击框架,模拟外部攻击者、恶意内部人员及高级持续性威胁的攻击方式。服务重点识别提示词注入、越狱、隐私窃取、智能体操控及业务逻辑漏洞,帮助企业在上线前发现问题并制定加固措施。
The NSFOCUS LLM Red Team Assessment Service examines security risks when large language models (LLMs) and AI agents are integrated into business applications.
It combines automated testing with expert-led manual assessment to evaluate models, application logic and underlying infrastructure from an attacker's perspective.
Informed by adversarial frameworks such as MITRE ATLAS, the service simulates techniques used by external attackers, malicious insiders and advanced persistent threats.
It focuses on prompt injection, jailbreaks, privacy theft, agent manipulation and business logic weaknesses, helping organizations identify issues and plan security improvements before deployment.
© 2026 NSFOCUS 绿盟科技 www.nsfocus.com All Rights Reserved . 京公网安备 11010802021605号 京ICP备14004349号 京ICP证110355号