Are Self-Driving Taxis Safe? What You Need to Know Before You Ride

On my recent trip to Austin, I was surprised at how many self-driving taxis there were. They were everywhere and are now popping up in more and more cities. If you haven’t seen one yet, you probably will soon. Companies like Waymo and Uber are already offering driverless rides in places like San Francisco and Phoenix. But even as the hype grows, one big question still lingers: Are self-driving taxis safe? That's what I wondered!

Some people love the idea — no distracted drivers, no one to have to make chit-chat with, no road rage — just smooth rides guided by tech. Others aren’t so sure, especially after hearing about glitches or accidents including smart cars. 

To get to the bottom of it, I did some research and asked some experts to chime in with their thoughts to break down how these vehicles work, what the safety data says, and what you should keep in mind before accepting a ride with no one in the driver’s seat.

Subscribe to our YouTube channel and learn how to protect your home, loved ones and, belongings.

To get a better idea of how self-driving cars work, I consulted with Eran Ofir, CEO of Imagry, a mapless autonomous AI software driving company. According to Ofir, despite the excitement surrounding autonomous taxis, there are still significant limitations in today’s robotaxi technologies that the public should be aware of.

“Most systems rely on high-definition (HD) maps and can only operate in pre-defined geo-fenced zones — typically limited to city centers or carefully mapped suburban areas,” Ofir told me. “As a result, these vehicles cannot navigate freely like human drivers. Crucially, they struggle with full trip coverage that spans both urban and highway segments — such as driving from downtown to an airport — which severely limits their usefulness for real-world transportation.”

Probably more importantly, most robotaxis currently roaming the streets in the U.S. (like Waymo, May Mobility, Mobileye, Uber, and Lyft) depend heavily on multi-modal sensor arrays, including lidar, radar, and multiple cameras. While this setup may appear to enhance safety, it often increases latency due to the heavy computational load required to fuse together data from all the sensors in real time. This can delay the vehicle’s response to fast-changing road conditions — ironically, posing a risk rather than mitigating one.

An outlier is Tesla’s robotaxi. Its platform takes a different approach. According to Ofir, it uses vision-only, mapless navigation, inspired by how humans drive — we see, interpret, and act. This makes the car rely solely on camera vision and real-time perception, enabling vehicles to adapt more naturally and swiftly to dynamic driving environments, with the potential for greater scalability and responsiveness.

Well, it’s hard to say, since human drivers have been around decades longer than autonomous vehicles. There's just more data for human driving and a huge lack of data for self-driving cars to make a good comparison, so far.

While self-driving cars remove human error, they aren’t without faults. Here’s what we know so far:

• As of July 25, 2025, the Department of Motor Vehicles has received 840 autonomous vehicle collision reports, according to the California Department of Transportation. 
• In April 2024, the Association for Computing Machinery warned to “not assume that fully automated vehicles will necessarily reduce road injuries and fatalities.” The paper stated that “Improved safety outcomes depend on appropriately regulating the safety engineering, testing, and ongoing performance of automated vehicles.”
• A study on Waymo cars found a 6.8 times lower rate in injury-related crashes than human driving after the company's cars had driven over 7 million miles.
• There were 40,901 people who died in car crashes in 2023, according to the National Highway Traffic Safety Administration. Overall, as of June 17, 2024, there have been 83 fatalities related to autonomous vehicle accidents. Remember, though, there are far fewer self-driving cars than human-driven.

So, generally, self-driving taxis are pretty safe, but what if you get the odd one out? Patrick Peterson, auto expert and team lead at Goodcar.com says there are a few things to watch out for while you’re riding around. 

If the car suddenly brakes for no reason, that’s a sign that something’s wrong. The vehicle should be driving smoothly, and any sudden stops could be due to sensor misreads or glitches, Peterson told me. Another big issue is “phantom braking,” where the car suddenly stops because it misinterprets something as an obstacle. In one study, 43% of autonomous vehicle crashes involved rear-end collisions. Those are often caused by this kind of braking. 

If the car fails to follow traffic rules, such as not yielding to pedestrians or not stopping at a red light, that’s a clear issue. “In San Francisco, there have been reports of Cruise robotaxis blocking emergency vehicles,” says Peterson. “These are signs that the system isn't fully understanding how to safely interact with human drivers or pedestrians. If you notice the car hesitating at intersections or making strange maneuvers, that could also be a red flag.”

If you think something is wrong with your autonomous taxi, use the emergency stop button (if it has one) and report any irregular behavior via the app or hotline connected to the car company, advises Ofir.

“Many robotaxis today are still in trial phases and can be monitored and even remotely controlled by human operators,” says Ofir. “In such cases, if something goes wrong, the vehicle can often be safely navigated to a stop or directed away from danger via remote human intervention. This serves as a safety net while the technology matures.”