Last year, Tesla defied its critics by boldly launching a robotaxi service that, by the end of the year, required no human supervision and was available to over 50 percent of the US population.

At least that’s what Tesla CEO Elon Musk told us would happen by the end of 2025. The reality, of course, was much different.

Tesla’s “robotaxi” service, as it stands today in Austin and San Francisco, is still not available to anyone who wants to use it. It is still supervised by an employee who sits in either the driver or front passenger seat with access to a “kill switch” if anything goes wrong. (There have been some unsupervised tests, but its unclear how many.) And last I checked, those two cities don’t comprise 50 percent of the population of the US.

For months now, Musk has been promising an “unsupervised” version of Tesla’s Full Self-Driving software that he claims will enable drivers to, among other things, use their phones while driving. The current version of FSD is a Level 2 “supervised” system, meaning drivers are required to stay focused on the road and be ready to take control of the vehicle when prompted. (In other words, not full self-driving.) And for years, Musk has been hyping a moment when Tesla’s vehicles will actually drive themselves — no supervision required.

But here we are, in the year 2026, and Tesla’s FSD is still very much a supervised system. Now Musk is saying it needs 10 billion miles in order to achieve this milestone. Right now, it has a little more than 7 billion miles, according to Tesla’s FSD dashboard. All of which begs the question why was Musk promising unsupervised FSD by the end 2025, when he knew Tesla had yet to achieve this new metric?

The answer is the goalpost is always shifting. In Tesla’s Master Plan Part Deux, released in 2016, Musk wrote that the company would likely need 6 billion miles “before true self-driving is approved by regulators.” Of course, there are no regulations standing in Musk’s way from launching a fully driverless, unsupervised system. Waymo seems to have all its permits in place.

My guess is that Tesla’s legal team is calling the shots here. As long as the driver remains responsible for what happens in the vehicle, Tesla can get away with the occasional wrongful death lawsuit. Unlike Waymo, Tesla doesn’t take responsibility for any incidents because it knows its vehicles aren’t fully autonomous. Why would it take responsibility for a Level 2 system? When Tesla owners have tried to force the company to accept liability for certain incidents, it has fought back in court — and often won.

But as soon as the system becomes unsupervised, the liability floodgates could open wide.