The surprising part isn't the theft

Someone stole yoga clothes in San Francisco. That is not, by itself, news. What is news is how they got away: by hailing a Waymo.

According to TechCrunch, a burglar used one of Waymo's driverless robotaxis as a getaway vehicle following a theft, and was not caught. The case is unremarkable as a crime story. As a data-governance story, it's more interesting.

What Waymo vehicles actually record

Waymo's robotaxis are equipped with an array of cameras, lidar (light detection and ranging sensors that build 3D maps of surroundings), and radar. These sensors run continuously while the vehicle is in operation — that's a core requirement for autonomous navigation, not an optional feature.

In principle, that means a Waymo vehicle that transported a suspect would have recorded the passenger's face, the pickup and drop-off locations, the route, and the timestamps. That's a more complete evidentiary record than most surveillance cameras provide.

In practice, whether any of that footage was preserved, requested, or usable in this case is not publicly known.

The retention and access question

Waymo has not published a detailed, public-facing policy explaining how long it retains passenger footage, under what legal standard it responds to law enforcement requests, or whether it proactively shares data when a crime is reported to have occurred in or around one of its vehicles.

That opacity matters. Ride-hail companies like Uber and Lyft have faced years of scrutiny over their law enforcement data-sharing practices, and both now publish transparency reports. Waymo, which is further along in autonomous deployment than any competitor in the U.S., has not matched that level of disclosure — at least not publicly.

This case doesn't resolve that question, but it makes it harder to ignore.

A rolling sensor platform with unclear rules

The broader issue is structural. Autonomous vehicles are, by design, among the most sensor-dense objects operating in public space. A single Waymo completing a shift in San Francisco generates a significant volume of data about the city's streets, pedestrians, and passengers.

That data has obvious value for navigation and safety. It also has obvious value for surveillance — by law enforcement, by litigants, and potentially by bad actors who might seek to access it. The question of who controls that data, under what rules, and with what oversight is one that regulators have been slow to address.

San Francisco, which has been the primary testing ground for Waymo's commercial expansion, has pushed back on the company on safety grounds. Whether city or state regulators will now press on data governance is an open question.

What we don't know

To be direct about the limits of what's reported here: it is not confirmed whether Waymo was contacted by police, whether footage was requested or provided, or whether any data gap contributed to the suspect's escape. TechCrunch's reporting flags that the incident "sheds new light" on Waymo's data practices, but the specifics of what Waymo did or did not do in this case have not been fully disclosed.

That uncertainty is itself part of the story. When a company operates a fleet of sensor-laden vehicles in a major city, the public arguably has an interest in knowing — in advance, not after an incident — what the rules are.