Is Global AI Governance Ready for Physical AI?

For the past few years, AI governance efforts have focused mostly on AI developments on bigger or smaller screens dominated by LLMs, disrupting traditional ways of interacting with AI, democratizing access to coding, text and image generation, benefiting companies and individuals while highlighting unprecedented risks. Meanwhile, physical AI or embodied AI developments were developing in the shadows. It appears that 2026 may be the breakthrough year when foundation models beyond LLMs will spread broadly in the physical environment, through robotics, advancements in autonomous vehicles, frontier models, next generation of AI powered devices for personal use, etc. Physical AI extends AI to systems that interact with and act in the physical world in various forms. The purpose of the panel is to discuss how advancements in global AI governance models and frameworks need to be applied and adapted to the challenges of physical AI. We intend to dive deeper into issues of data governance through the physical AI lifecycle, from training data sets and their representative character to bias minimization, application of PETs, data collection at deployment stage and transparency challenges. Panelists will discuss case studies of current developments related to humanoid robots and industrial processes. They will touch upon the framework of stakeholders in the physical AI supply chain. The panelists represent a global approach including South Korea, Europe, and China, and will be moderated by an AI governance and privacy leader with a worldwide footprint and a growing presence of physical AI models in the operations of the group's tech companies.