ETSI Secures AI: New Standards for Data Centers & Cloud

ETSI has released a new technical specification, TS 104 033, which defines security requirements for AI computing platforms. This specification creates a security framework for platforms that host AI applications in data centers and edge computing environments. The framework covers security functions, platform components, interfaces, and services. It aims to protect AI models, datasets, training processes, and inference workloads. Scott Cadzow, Chair of the ETSI Technical Committee Securing AI, states this is a significant step in establishing concrete security requirements for these platforms. The specification organizes security requirements into categories like identity management, data protection, and incident response. It mandates features such as secure boot mechanisms and prohibits remote access to root accounts. The platform must also protect data during transmission and support backup and recovery. What's important is that this specification addresses risks unique to AI environments. This includes protecting AI models and datasets, and ensuring isolation between users on shared AI accelerators like GPUs. The platform should also detect attacks on AI inference processes and securely store AI-related logs. The bottom line is these new requirements establish a foundation for secure and trustworthy AI operations.