For decades, the law treated computers as instruments—sophisticated typewriters, ledgers, or calculators. Even as machine learning advanced, software remained a tool directed by human hands.
 
That assumption breaks down when systems begin to act: to set sub-goals, enter commitments, execute transactions, and negotiate with other autonomous agents across borders and jurisdictions.
 
In this opening keynote, Dr. Nicklas Lundblad examines the friction between agentic AI and fundamental legal doctrine. What does "agency" mean when intent is statistical? Where do responsibility, liability, and duty of care fall when no person directed the specific outcome—and no single engineer can fully reconstruct it?
 
Drawing on his years inside frontier AI labs and his work in strategic foresight, Lundblad looks past compliance checklists to a harder question: how legal categories, institutions, and reasoning must adapt when agency is no longer uniquely human.