I published a technical note describing a protocol where AI systems are structurally prevented from making decisions.

The design enforces: - non-decision by design - explicit stop conditions - human responsibility at every step - traceability of interactions

The goal is to make misuse structurally impossible rather than policy-based.

DOI: 10.5281/zenodo.18100154