The AI Standards Hub – Global Summit 2025, organized by the UK's AI Standards Hub, is the inaugural global summit on AI standardization, held in London from March 17 to 18, 2025, with online streaming available. Co-hosted by the Organisation for Economic Co-operation and Development (OECD), the Office of the United Nations High Commissioner for Human Rights (OHCHR), and the Partnership on AI, the summit aims to explore the critical role of standards in AI governance, promoting global inclusiveness and collaboration.
The first day focuses on the current state of AI standardization, discussing how standards can facilitate international regulatory interoperability, enhance stakeholder diversity, and build an effective AI assurance ecosystem. The second day centers on key issues in the governance of foundation models, examining opportunities for alignment and closer collaboration between the AI safety and standardization communities. The following will present the content of two of the sessions:
1. The Lay of the Land – Technical Measurement Standards for AI This session delves into the application and challenges of technical measurement standards in the AI field, with experts from NIST, KRISS, NPL, and others sharing their experiences. The report points out that AI systems often lack transparency in data sources and decision-making logic. Metrology institutions focus on enhancing data traceability and decision-making transparency. Moreover, research in metrological semantics helps establish structured classifications of AI-related terminology, promoting consistent understanding across countries and industries. In terms of regulatory implementation, metrology institutions provide reliable assessment and certification methods, assisting enterprises in swiftly achieving compliance. However, current evaluation methods for AI systems are still developing, especially in measuring aspects like bias and fairness in socio-technical systems. Moving forward, metrology institutions will continue to integrate international experiences and industry needs to advance the systematic and trustworthy standardization of AI.
2. The Lay of the Land – Global Standardization for AI This session features experts from international standardization organizations, including Silvio Dulinsky (ISO Deputy Secretary-General) and Philippe Metzger (IEC Secretary-General). The report highlights the increasing importance of establishing consistent and trustworthy standards as AI technologies are widely applied in critical sectors like healthcare and transportation. The U.S. NIST's AI Risk Management Framework provides guidance on assessing AI systems' trustworthiness characteristics, while Germany's PTB focuses on developing quantifiable explainability assessment tools. Additionally, the EU's AI Act emphasizes that high-risk AI systems must comply with specific standards, with metrology institutions playing a crucial role. National metrology institutes also stress the importance of creating shared taxonomies and semantics to facilitate international standardization dialogues.
