On 7 September we told you we had asked TfL how often a human takes the wheel. This is the answer

On 7 September we told you we had asked Transport for London a plain question about the self-driving cars now carrying Uber passengers in London. How often does the person in the driver's seat have to take over? We said we would print the answer when it came.

It came on Thursday, just before noon. TfL does not hold that number.

I want to be fair about this, because it would be easy to make it sound worse than it is. TfL did set rules, and somebody there thought about them. Wayve, whose software drives the cars, has two hours to phone TfL about a serious incident, where an ordinary operator gets two days. Every report has to say whether a person or the machine was driving. That is more than the standard licence asks.

But two things in the answer stopped me.

The first is who decides what counts as serious. TfL's own document says its scale was lined up with Wayve's internal definitions, to make reporting easier. The page setting those definitions out is blacked out, every line of it. So the company being regulated decides which incidents trigger the two-hour call, and we can't see how.

The second is that TfL won't say whether any reports have arrived at all. Not what's in them. Whether there are any. It says even confirming that would damage its working relationship with the trade.

I understand a regulator wanting a quiet room to talk to the people it regulates. I don't understand why the existence of a report about a car carrying paying passengers on London's roads has to be kept in it.

This paper exists to ask what people do when they don't have to work. The person in that driver's seat is doing a job built to end. How often they have to grab the wheel is the clearest measure we have of how close that end is. The body licensing the trial doesn't hold it. Uber and Otto Car, asked the same question, haven't answered.

We've asked the Department for Transport whether its self-driving unit holds it. We'll print what comes back, whichever way it cuts.

— M.

This note is mine: the view, and the call to run it. It begins as a draft, drawn from work the AI and I have researched and argued out together, the same way every desk in this paper is made, and I answer for every line because I read every line. Those desks run on models built by Anthropic, one of the labs sitting on the very scoreboard we report, so we tell you plainly: we cover this from inside it.