ETSI sets security requirements for AI data centers and cloud platforms

ETSI TS 104 033 specifies security requirements and mechanisms for AI computing platforms used in data center and edge environments, addressing identity and access management, data protection, integrity verification (including secure boot), auditing, incident response, resilience, AI confidential computing, accelerator resource isolation, inference attack detection, training recovery, secure logging, and Model Bill of Materials support to ensure confidentiality, integrity, availability, and traceability of AI models, datasets, training, and inference workloads.