Autonomous Drones: Who Answers When the Machine Decides?

A delivery drone swerves to avoid a flock of birds, crosses an unplanned air corridor and damages a roof. Nobody was at the controls: the decision was made in flight, by software trained on millions of scenarios. Whose fault is it? The drone manufacturer's, the navigation software vendor's, the operator who approved the mission, the regulator who certified the whole system?

The question is no longer speculative. Autonomous drones inspect power lines, deliver medicine to isolated areas, support mountain rescue teams. Every successful mission further normalises delegating decisions to the machine — and every incident exposes the legal vacuum that comes with it.

Traceability as the First Requirement

Legal scholars are converging on one central demand: traceability. An autonomous aircraft must be able to reconstruct, second by second, what it perceived, what it inferred and why it chose as it did. Without that interpretable black box, no chain of liability holds — and insurers refuse to play.

Behind the drone case looms a precedent for all mobile robotics, from shuttles to agricultural robots. The answer we settle on — diluting responsibility or organising it — will determine how much trust these machines deserve.