AI Security Is an Engineering Problem — How to Solve It at Every Layer of the Agent Stack
中文摘要
人工智能安全是一个工程问题。这意味着需要明确的安全需求、可执行的控制措施、指定的负责人以及证明保护措施有效的证据。随着人工智能能力的提升，业界必须加快安全工程的发展，扩大对防御工具的访问范围，并更快地分享有效的做法。技术在变化，安全基础却始终如一。互联网和云计算改变了软件的运行方式，[...]
英文原文
AI security is an engineering problem. That means defined security requirements, enforceable controls, named owners and evidence that protections work. As AI becomes more capable, the industry must accelerate security engineering, broaden access to defensive tools and share what works faster. Technology Changes, Security Fundamentals Endure The internet and cloud computing changed how software operates, […]