People Are Back-Flipping Off of Waymo’s Robotaxis

In a bizarre (or hilarious) late-night episode that underscored public unease with autonomous vehicles, several men climbed onto stalled Waymo robotaxis in San Francisco’s Marina District and began attacking them.

They then started sitting and climbing on them, and are one point began doing back flips off the driverless cars while a crowd cheered.

City police eventually cleared the scene, but the incident highlights growing tensions over deployments of robotaxis in urban areas.

So what happened?

Roughly around 2 a.m. last Sunday, ABC7 Bay Area captured unsettling footage of three Waymo vehicles immobilized at the intersection of Fillmore and Greenwich streets.

Dozens gathered around as individuals sat atop the cars and gestured at their sensors. ABC7 reported that no visible damage was done and no passengers were inside.

One onlooker, captured in the video, was seen doing a back flip off of the robotaxi.

Selika Josiah Talbott, a veteran federal advisor in autonomous vehicle regulation, called the behavior “shocking and horrifying,” warning that such stunts teach AI systems that crowds are aggressive, skewing their behavior in future deployments.

“It’s also still dangerous. The leaps that these kids were doing … had it been their head hitting the ground, it’s just incredibly dangerous and illegal,” Talbott told ABC7. She urged that “law enforcement must take these incidents seriously … at least in the beginning to send a message.”

Waymo and the Bay Area have a long history

The incident isn’t the first sign of friction.

The Washington Post reported that in 2024 alone, Waymo vehicles received 589 parking tickets in San Francisco for obstructing traffic, violating street-cleaning rules, and impeding emergency responders. In one case, Waymo robotaxes caused over two hours of transit delays, including blocking fire trucks responding to emergencies.

It’s a trend not limited to minor violations. The Guardian reported robotaxis were vandalized by a mob—painted, set ablaze, or damaged with traffic cones—amid a protest against autonomous vehicles in crowded neighborhoods. These acts illustrate the deep frustration among residents who feel their safety and street equity are being compromised.

“The vehicle was not transporting any riders and no injuries have been reported,” Waymo said in a statement at the time. “We are working closely with local safety officials to respond to the situation.”

From tech showcase to cultural flashpoint

San Francisco was an early adopter of robotaxis, but increasingly the city has become a stage for real-world tests of how the technology actually interacts with people and the places they live.

Waymo, now part of Alphabet, has rolled out in multiple urban areas such as Phoenix, Los Angeles, and Austin, and has over 200,000 paid rides per week by early 2025. But they are far from being ubiquitous and the novelty surrounding them.

The Marina incident highlights broader societal questions: Do robotaxis, still a novelty, provoke unintended psychological responses from people, especially late at night? Could their hesitation in complex urban settings be misinterpreted, leading to frustration, confrontation, or worse? In one case, the mob itself took over.

“That was when it went wild,” Michael Vandi, told Reuters, about the incident involving arson. “There were two groups of people, folks who encouraged it and others who were just shocked and started filming. No one stood up. There wasn’t anything you could do to stand up to dozens of people.”

What comes next for regulators?

Autonomous vehicle companies and regulators must grapple with more than collision avoidance and mapping, they need to address human reactions in public spaces.

The California DMV has proposed the authority to issue citations directly to self-driving firms beginning in 2026, signaling regulatory urgency. Cities may also impose mandatory “social impact” testing or require in-vehicle safety drivers in specific zones.

Meanwhile, vehicle designers must consider behavioral cues and cameras that allow robotaxis to signal intentions to nearby pedestrians, or to figure out how to get out of situations like being attacked while stalled.

Engineers are experimenting with tools like that, perhaps via shifting lights or audible signals, or something as simple as mirroring human eye contact in traffic. That sort of help might have come in handy during this week’s mini Waymo riot, a bystander said, because even the cars weren’t sure what to do next.

“There was an officer that appeared and eventually sort of shooed everyone to the side so that the vehicle to begin to move,” Talbott told the news station. “Even the vehicle’s hesitancy as it began to operate again was because of the intrusiveness of humans.”

So what now?

What began as an experiment in urban transportation is fast becoming a cultural flashpoint, where human expectations collide with cutting edge technology.

But without a driver to guide or defend them, how safe are both the people who use them and the cars themselves if they run into hostile onlookers and can’t leave immediately? That may be a question that even the companies that make driverless cars can’t answer.

“What is becoming abundantly clear is that AV technology is not as sophisticated as the industry would like us to believe,” California state Senator Dave Cortese told Reuters.

The Marina spectacle isn’t just a viral oddity. It’s a wake-up call. Waymo and the industry need to navigate not just streets, but societal terrain where public confidence must be earned one respectful, safe interaction at a time.

“We are seeing people reaching a boiling point over tech that they do not want and does not make their lives better,” Missy Cummings, director of the George Mason University Autonomy and Robotics center and a former adviser to U.S. traffic safety regulators, told Reuters.