Rubrik 推出 AI 代码安全工具 在隔离环境中替代码库做红队测试
网络安全公司 Rubrik 10 月 9 日宣布扩大 Project Hourglass 合作联盟，并发布由 Anthropic 的 Claude Mythos 5 驱动的代码安全工具 Rubrik Code Guardian，开放私人预览。AHEAD、Trace3 和 WWT 加入联盟，帮助客户在 AI 编程加速的同时守住代码安全。
Code Guardian 不直接碰生产环境，而是在隔离（air-gapped）的代码库克隆中做红队测试，跨文件、服务、身份和云边界寻找可以连成的攻击链，再按可利用性和业务影响排序，并配合 Rubrik 的代码恢复能力。Rubrik 称，86% 的网络安全与 IT 负责人预计，未来一年 AI 智能体的扩张速度会超过组织自身的安全护栏。该工具目前仅向选定的设计伙伴开放，尚未正式全面上市。（资料来源：Rubrik 官方新闻稿，2026 年 10 月 9 日）
Rubrik Unveils AI-Powered Code Guardian to Red-Team Software in Air-Gapped Environments
Rubrik on October 9 expanded its Project Hourglass alliance and introduced Rubrik Code Guardian, an AI code-security tool powered by Anthropic's Claude Mythos 5, now in private preview. AHEAD, Trace3 and WWT joined the alliance as engineering teams adopt AI coding tools at speed.
Instead of testing live production systems, Code Guardian works on cloned, air-gapped copies of customer code repositories. It reasons across files, services, identity roles and cloud boundaries to uncover chained vulnerabilities that conventional static tools may miss, verifies exploitability, prioritizes findings by business impact and pairs the analysis with Rubrik's codebase recovery capabilities. Rubrik says 86 percent of cybersecurity and IT leaders expect AI agents to proliferate faster than their organizations' security guardrails within the next year. Code Guardian is not yet generally available. (Source: Rubrik official release, October 9, 2026)
图片来源：图片来源：Openverse · Visual Content（CC BY）