Sophia Antipolis, France, 2 June 2026
ETSI is pleased to announce the release of a new technical specification defining security requirements for Artificial Intelligence (AI) computing platforms. The ETSI TS 104 033 specification provides a systematic security baseline for the computing platforms that host AI model training and inference.
In AI systems, the computing platform serves as a critical infrastructure layer, providing the execution environment and associated resources for AI services. It enables a “secure by default” platform for running AI models. As AI systems are increasingly deployed across a wide range of use cases, the security of the computing platform has become a key element in safeguarding the entire AI supply chain.
The new ETSI specification addresses this critical need by defining security requirements and associated functions for AI computing platforms, helping to mitigate threats to the platform and its assets (such as models and data). It also specifies the security components to be implemented within the platform, as well as their interfaces.
As stated by Scott Cadzow, Chair of the ETSI Technical Committee Securing AI, which is responsible for the specification: “This work builds on the AI computing platform security framework we have previously developed and marks a significant step forward in establishing concrete and actionable security requirements for the platform itself”.
- The technical specification covers the following aspects:
Security requirements and functions – Defines the mandatory security requirements and associated functions that AI computing platforms should meet to mitigate security threats. - Security components and service interfaces – Specifies the security components and service interfaces recommended for implementing these requirements and functions.
- Asset Protection – Addresses the protection of AI models and data in dynamic states (in use/in transit) and static state (at rest), helping prevent threats such as model extraction and data leakage.
- Alignment with ETSI EN 304 223 – The security requirements align with the core principles and five lifecycle phases defined in ETSI EN 304 223 (secure design, development, deployment, maintenance, and decommissioning), contributing to end-to-end security from platform to system to model.
The release of this technical specification provides a clear security baseline for stakeholders such as AI computing platform designers, supporting compliance with regulatory requirements, including those set out in the EU AI Act.
About ETSI
ETSI is one of only three bodies officially recognised by the European Union as a European Standards Organisation (ESO). It is an independent, not-for-profit body dedicated to ICT standardisation. With over 900 member organisations from more than 60 countries across five continents, ETSI offers an open and inclusive environment for members representing industry, research institutions, academia, governments, and public bodies. ETSI supports the timely development, ratification, and testing of globally applicable standards for ICT-enabled systems, applications, and services across all sectors.
For more information, please visit: https://www.etsi.org/
Contact:
Ms Claire Boyer
Cell: +33 (0)6 30 48 58 95
Email: [email protected]
