研究人员在流行的开源框架LangChain中发现一个关键漏洞，该框架广泛用于驱动大型语言模型（LLM）Agent。该漏洞编号为CVE-2025-68664，CVSS评分高达9.3分，攻击者可能利用该漏洞提取敏感环境变量或触发非预期的系统操作。 漏洞根源：数据序列化缺陷 该漏洞源于 ...

人工智能安全初创公司Cyata Security Ltd.今天发布报告，详细介绍了最近在langchain-core中发现的一个严重漏洞。langchain-core是基于LangChain智能体背后的基础库，在人工智能生产环境中被广泛使用。 这个漏洞被追踪为CVE-2025-68664，绰号"LangGrinch"，通用漏洞评分系统得分 ...

Would you trust an AI agent to run unverified code on your system? For developers and AI practitioners, this question isn’t just hypothetical—it’s a critical challenge. The risks of executing ...

What if you could build an AI system that not only understands your needs but also intelligently decides how to act on them? Imagine a virtual assistant that doesn’t just answer questions but ...