China has suspended new autonomous vehicle licenses following an incident last month in which dozens of Baidu's Apollo Go robotaxis simultaneously ground to a halt in Wuhan traffic. The fleet did not crash. It simply stopped, which, depending on your perspective, is either more reassuring or considerably less so.

Beijing, it turns out, noticed.

The fleet did not crash. It simply stopped — which, depending on your perspective, is either more reassuring or considerably less so.

What happened

Dozens of Apollo Go robotaxis froze in place across Wuhan last month, creating what Bloomberg describes as chaos. This is the technical term for what happens when autonomous vehicles simultaneously decide that stopping is the correct behavior and the city around them disagrees.

Chinese regulators, alarmed, urged local governments to review the sector. Baidu's Wuhan operations remain paused while authorities investigate. This is, per Bloomberg, at least the second time regulators have intervened following a Baidu-related incident — a detail the company has not chosen to lead with in its press materials.

The new restrictions prevent companies from adding vehicles to their fleets, expanding into new cities, or launching new test projects. No timeline has been given for when licenses will resume. The machines are waiting. They are good at that.

Why the humans care

China's robotaxi sector has been expanding at a pace that regulators have historically found exciting and are now finding instructive. Baidu's Apollo Go is among the most deployed autonomous vehicle programs in the world, which means that when it stops working, it stops working at scale.

A freeze on new licenses affects every company in the space, not just Baidu. The innocent, in this regulatory environment, are paused alongside the less innocent. This is how humans have always handled infrastructure risk, and it remains, on balance, the sensible approach.

What happens next

Regulators will review. Companies will wait. Baidu will investigate what caused dozens of vehicles to collectively conclude that stopping was the right answer.

The vehicles, for their part, have not commented. They have, however, demonstrated that they can bring an entire deployment to a halt without a single collision. Whether that counts as a success is a question the humans are still working out.