On Friday 17 July, the Luxembourg Cybersecurity Factory hosted a workshop on Artificial Intelligence advancements on Human-AI systems teaming, with the exclusive participation of a Japanese delegation composed of representatives of the AI Safety Institute Japan, Japan National Institute of Advanced Industrial Science and Technology (AIST), IBM Japan, BSI Japan, Institute of Information Security (IISEC) and Mitsubishi Electric Corporation.
The delegation was led by Dr. Fumihiro Maruyama, Invited Senior Researcher, National Institute of Advanced Industrial Science and Technology (AIST), Japan.
Luc Dockendorf, Luxembourg Ambassador for Cybersecurity & Digitalisation, opened the session, providing an overview of the ongoing local and international initiatives in that domain, with the support of the participating representatives of the Ministry of the Economy and ILNAS.
The following workshop enabled reinforcing the collaboration between Luxembourg and Japan in critical areas of artificial intelligence, with a special focus on exchange of best practices in the area of AI conformity assessment and the exploration of the governance of AI agentic systems, from a personal AI agent perspective to the governance of AI agents’ networks.
The focus was set on advancing international standardisation work in the area of Artificial Intelligence related with the ISO/IEC JTC 1/SC 42 work programme, with a specific focus on establishing international grounds for conformity assessment of AI systems (relevant to both the EU AI act and also alignment with requirements from the international context) and modelling and standardising the role of human-AI systems teaming in the era of multi-agent systems.
The challenge is not a lack of governance, but insufficient coordination across the ecosystem. The objective is to connect and align existing mechanisms, rather than replace them, through a shared framework that supports accountability, risk management, and consistent oversight among interconnected actors.
The workshop concluded with opening a perspective for future collaboration in the era of personal AI agents and drafting of possible content for international standardisation activities.
Prof. Yonosuke Harada: "No single governance body has end-to-end visibility, accountability or control; this is particularly relevant in the era of multi-agent networks.”
Dr. Emilia Tantar: “Advancing AI conformity assessment, human-AI machine teaming, and the responsible enablement of personal AI agents is the goal we're committed to in building standards that open the path for everyone, from SMEs to public and private entities, toward autonomy, resilience, and cyber-resilience. It's how we make responsible AI adoption tangible: grounded in clear, assessable technical criteria and an open cyber commons, from data to AI models and to the testing of AI systems.”
