Screenshot from Youtube’s Black Tesla of Autopilot demanding the driver retake the wheel.

Tesla drivers are suing the company for making them “beta testers of half-baked software that renders Tesla vehicles dangerous,” as Bloomberg reports. That’s the strongest condemnation of the sophisticated but sometimes controversial Autopilot system we’ve seen yet.

The complaint (read the full text here) was filed yesterday in the U.S. District Court in San Jose, California, seeking class-action status for the owners of 47,000 Model S and Model Xs sold in 2016 and 2017. The problem, according to the complaint, is that Autopilot is dangerous as hell and it’s the customers who are working out the bugs, as Bloomberg notes:

Consumers allege their cars — purchased for $81,000 to $113,000 — veer off lanes while often “lurching, slamming on the brakes for no reason, and failing to slow or stop when approaching other vehicles” when Autopilot is activated.

Tesla did officially respond to the suit, and called it “disingenuous,” “inaccurate” and “sensationalist,” for saying that Tesla didn’t make it 100% clear what Autopilot could or could not do and how or how not it should be used:

This lawsuit is a disingenuous attempt to secure attorney’s fees posing as a legitimate legal action, which is evidenced by the fact that the suit misrepresents many facts. Many of the features this suit claims are “unavailable” are in fact available, with more updates coming every month. We have always been transparent about the fact that Enhanced Autopilot software is a product that would roll out incrementally over time, and that features would continue to be introduced as validation is completed, subject to regulatory approval.

Furthermore, we have never claimed our vehicles already have functional “full self-driving capability”, as our website has stated in plain English for all potential customers that “it is not possible to know exactly when each element of the functionality described above will be available, as this is highly dependent on local regulatory approval.” The inaccurate and sensationalistic view of our technology put forth by this group is exactly the kind of misinformation that threatens to harm consumer safety.

Tesla, for better or worse, has always said Autopilot is a beta. The correctness, however, of testing a beta on actual human customers is debatable. Certainly it’s repeatedly getting Tesla into hot water with Autopilot.

Advertisement

The lawsuit focuses on the second-generation update of Autopilot that was debuted last fall but Autopilot’s sketchiness came up almost immediately after the tech debuted in 2015. One driver swore his car tried to kill him when it veered out of its lane on a twisting back road.

Tesla argued that one should not “abdicate responsibility” with Autopilot on, and should use it on clear highways with your hands on the wheel.

Autopilot has also been involved in at least one death when a driver with Autopilot engaged crashed into a semi truck pulling out across a divided highway. It’s not clear if the driver was misusing Autopilot, not paying as much attention as he should have or if the Autopilot system had a dangerous blind spot. There have been many more complaints about Autopilot, often likely user error. Tesla’s position in those has been consistent.

Tesla then dialed back its Autopilot capabilities only to open them back up not long ago with a recent update.

The problem, it may turn out, is not in the technology itself, but relying on Tesla owners to use it responsibly. What works in the minds space of Silicon Valley doesn’t always translate well to the rest of the world.