The National Highway Traffic Safety Administration’s recently announced investigation into self-driving company Waymo is just the latest of many such probes.

Since October, NHTSA has opened investigations into autonomous vehicle systems operated by Tesla, Ford, Zoox, and Cruise, suggesting a significant shift in approach to the nascent world of self-driving vehicles.

The probes have stepped up in number as the incidents involving self-driving vehicles have increased. Social media’s ubiquity has also meant that shocking videos of vehicles losing control, or disregarding the rules of the road, have been shared widely—potentially prompting NHTSA to step in.

“Establishing potential technology failures of autonomous vehicles is essential both for future improvements and for establishing public trust in the technology,” says Richard Lawton, head of communications at DriveElectric, an EV firm. “It is only right that regulations and standards are sufficiently robust to ensure this technology is safe.”

The investigation into Waymo—which is owned by Google parent Alphabet—involves 22 separate incidents in which the robotaxis either “exhibited driving behavior that potentially violated traffic safety laws” or were the only operating vehicle in a collision, per the NHTSA investigation document. NHTSA added that it hoped to “evaluate the ADS’s [automated driving system’s] performance in detecting and responding to traffic control devices and in avoiding collisions with stationary and semi-stationary objects and vehicles.” Waymo, for its part, said in a statement that it would cooperate with the investigation, and pointed to “tens of millions of autonomous miles driven” without issue, alongside its “demonstrated commitment to safety transparency.”

Jack Stilgoe, a science and technology professor at University College London, says that Waymo and other self-driving technologies would struggle to pass a traditional human driving test—not necessarily because they can’t drive, but because the test is not designed for them. And Waymo’s callout to the millions of miles driven without incident is a highlight of that disparity. Cruise did something similar last year, publishing research in collaboration with two universities that suggested self-driving cars it operated had less than half the number of crashes per million miles driven than human rideshare drivers.

Stilgoe argues that such a claim manages to couch what may be a bigger issue. “With driving, it’s not the total number of miles that matters,” he says. “It’s what happens in the tiny minority of circumstances where you’re called upon to make a difficult decision. That’s where safety is defined.”

And that’s the point at which NHTSA appears to be probing companies developing these systems more intensely. For years, regulators have stood back and allowed the industry to develop autonomous driving tech with little oversight. But as incidents start to rise, both in absolute numbers and seriousness, regulators see the need to step in.

“Depressingly, this is how regulation tends to happen,” Stilgoe says. “Regulators typically operate in this sort of wait-and-see mode, particularly in the United States.”

Phil Koopman, a professor at Carnegie Mellon University who has been working on self-driving car safety for more than 25 years, explains that the U.S. relies on two main regulatory gatekeepers: NHTSA, which plays a reactive role, and lawsuits. “Due to the number of vehicles on the road, most of these have been directed to Tesla,” Koopman says. “For a while Tesla was winning all the lawsuits. But given the recent settlement we might see other pressure from pending lawsuits for Tesla and other self-driving or robotaxi companies to up their safety game.”

It seems unlikely that the self-driving revolution will hit a brick wall in the face of regulation, Stilgoe says; rather, regulators will accede to softening their stance in order not to get in the way of technological advancement. “What I would be concerned about,” he says, “would be if regulators rolled over and bought the argument about the improved safety of self-driving vehicles and changed their rules to suit the technology.”