[ivory-search id="3463" title="Default Search Form"]
Search
Search
Type below to search:
Room V
Workshop
In person
LeadersGoldDiscovery

AI foundation model evaluation and standards

    • Date
    8 July 2026
    Timeframe
    14:30 - 18:00 CEST
    Duration
    3h 30 minutes

      AI Foundation models, including large language models (LLMs), multimodal models, and the rapidly emerging class of AI agents built upon them, are reshaping how industries innovate, how public services are delivered, and how societies address shared challenges from healthcare to climate action. As these models grow in scale, autonomy, and deployment, a robust and internationally harmonized approach to their evaluation and standardization has become essential to ensure their safe, trustworthy, and beneficial use worldwide. 

      Currently, the international community has yet to converge on unified evaluation indicator systems, cross-border mutual recognition mechanisms, or collaborative pathways for open benchmarking of foundation models. There is therefore an urgent need to foster convergence on evaluation methodologies, promote interoperability of benchmarks, and advance harmonized standards that can guide the responsible development of both foundation models and AI agents. Building on the foundational standardization work of ITU-T Study Group 21, this workshop gathers stakeholders from across the standardization, research, and industry communities to build consensus on evaluation approaches for both foundation models and AI agents. 

       

      Schedule

      This ceremony marks the launch of an international initiative calling on global organizations to jointly build foundation model evaluation capabilities, share benchmarking resources and best practices, and advance responsible AI development. Grounded in ITU-T's standardization work on foundation model benchmarking, the initiative aims to establish a shared action framework for trustworthy, controllable, and beneficial foundation model evaluation.

      Speaker(s)

      This session examines the state of the art in foundation model evaluation through three keynote presentations. Topics include global benchmarking infrastructure, the role of open-source ecosystems in advancing evaluation methodologies, and industry perspectives on deploying evaluation frameworks at scale.

      Moderator(s):

      This panel explores how evaluation frameworks can serve as a bridge between technological progress and responsible deployment. We examine whether current capability benchmarks fully capture the broader picture of AI advancement—not just performance gains, but a deeper understanding of societal implications. Panelists will discuss how multilateral cooperation can establish shared norms for AI assessment, enabling technological development to align naturally with humanity's long-term interests as it continues to advance.

      Panelists:
      Moderator(s):

      This session shifts focus from capability to governance, examining how international standardization can provide the institutional backbone for trustworthy AI evaluation at global scale. Speakers will address standardization roadmaps, independent evaluation approaches, and the technological frontiers shaping future standardizations needs.

      Moderator(s):

      This panel convenes representatives from major standards organizations to discuss how the international standards ecosystem can support coherent, interoperable AI evaluation frameworks. The discussion will explore mechanisms for mutual recognition, inclusion of developing countries, and coordination across ITU, IEC, and other bodies.

      Panelists:
      Moderator(s):
      Share this session with your network
        Add to Calendar
        In partnership with
        Related stories

        India Mobile Congress 2026: Connecting AI innovation, skills and real-world...

        (Photo: India Mobile Congress) India Mobile Congress 2026 brought together young innovators, researchers, policymakers and international organisations in New Delhi...
        AI for Good | Global Summit | Robotics for Good

        Trust isn’t coded – it must be designed into the...

        Artificial intelligence is advancing at extraordinary speed. Yet the pace at which organizations and nations adopt it will increasingly be...
        AI for Good | Global Summit

        Why we were at AI for Good: grounding artificial intelligence...

        Ask anyone building or using AI tools for research what their biggest concerns are, and hallucinations and citation problems come...
        AI for Good | Global Summit

        Are you sure you want to remove this speaker?