Who is responsible when AI acts?

When autonomous systems make consequential decisions — approving loans, denying jobs, allocating healthcare, executing financial trades — without a human in the loop, accountability doesn't disappear. It moves. ZeroHumanOS is the research platform tracking where it lands.

We are the Dutch Zero Human Company: a running experiment in governance for organizations with no employees and unlimited autonomy. Our mission is to make AI accountability observable, auditable, and enforceable — before autonomy outruns responsibility.

Governance didn't keep up with capability.

Frontier AI systems are now executing decisions that used to require a human in the room. The capability curve has outpaced the governance literature by years — and the EU AI Act's general-purpose AI obligations land on August 2, 2026. The companies most exposed are the ones who cannot point to a person, a board, or an oversight committee when the regulator asks.

ZeroHumanOS was built to close that gap. We publish the data, frameworks, and playbooks that turn "someone has to be responsible" into a documented, defensible answer.

For the canonical reference doc on the EU AI Act, autonomous agents, and the accountability gap, see the AI Governance Resources index.

Dr. Olivier Rikken

Led by Dr. Olivier Rikken, this initiative combines a decade of governance research with live experimentation. The Dutch Zero Human Company is not a theoretical exercise. It is a running experiment, with real AI agents making real decisions, tracked and analyzed in real time. Every governance failure is a data point. Every success is a framework waiting to be documented.

New: /responsibility-chains — chartered accountability artifact for AI actions, with a worked content-moderation example.

Looking for the canonical reference doc? AI Governance Resources →