Robotaxis Save Lives… Then Do This

Waymo’s self-driving cars have logged 170 million miles and cut serious injury crashes by 92% compared to human drivers — yet the same vehicles have run red lights, driven into floods, and blocked an ambulance during a mass shooting response.

At a Glance

Waymo reports a 92% drop in serious injury crashes across 170 million fully autonomous miles — but the safety data is self-reported and has not been independently verified.
CNN found Waymo vehicles ran red lights, crossed double yellow lines, and drove into flooded streets with passengers inside.
The National Highway Traffic Safety Administration (NHTSA) forced a recall of nearly 4,000 Waymo robotaxis after a software flaw caused them to drive into closed construction zones.
Critics question whether blocking self-driving car expansion — despite its real flaws — may cost more lives than it saves, given how deadly human drivers already are.

The Safety Numbers Are Hard to Ignore

Waymo says its driverless cars have now driven more than 170 million miles without a human at the wheel. That is equal to more than six human lifetimes of driving every single week. According to Waymo’s own safety data, its vehicles cause 92% fewer crashes with serious or fatal injuries than human drivers. They also cause 82% fewer crashes with any injuries at all. At the company’s current pace of 4 million miles per week, Waymo estimates it prevents roughly one serious injury crash every eight days.

These numbers are striking. About 40,000 Americans die in car crashes every year. Human error causes the vast majority of those deaths. If self-driving cars truly perform as Waymo claims, slowing their rollout is not a neutral choice — it is a decision that costs lives. Some doctors and economists have made exactly that argument in major newspapers, calling delays a “public health failure.”

Real Problems on Real Roads

The safety record is not clean, though. A CNN review of public records found Waymo vehicles in Austin, Texas ran red lights, crossed double yellow lines, and drove on the wrong side of the road — hundreds of times. In San Antonio and Austin, Waymo cars drove into flooded streets during flash floods, trapping passengers inside until the company issued a voluntary software fix. During a mass shooting response in Austin, a Waymo vehicle blocked an ambulance for about two minutes. Paramedics had to physically move the car.

In January 2026, NHTSA confirmed that a Waymo vehicle struck a child near an elementary school in Santa Monica, California. The car slowed from 17 miles per hour to below 6 miles per hour before impact. The child suffered minor injuries. Separately, NHTSA ordered a recall of nearly 4,000 Waymo robotaxis after a software defect caused vehicles to ignore construction zone closures and drive into blocked-off areas in San Francisco and Phoenix.

Questions About the Data Itself

Safety advocates have raised concerns about how Waymo counts crashes. According to an analysis reported by The Verge, 45% of Waymo’s reported crashes involved no passengers at all, and nearly 80% were rear-end collisions — compared to about 30% for human drivers nationally. That unusual pattern raises fair questions about whether the data reflects real-world driving risks or a narrower set of conditions. No government agency or independent university has yet audited Waymo’s raw crash data to confirm its 92% safety claim.

There is also a transparency gap on the human side of the operation. CNN found that Waymo employs 70 remote operators who can suggest maneuvers when the car faces a confusing situation. About half of those operators are based in the Philippines and do not hold U.S. driver’s licenses. Waymo markets its vehicles as fully autonomous. Critics say that description needs more scrutiny. Meanwhile, 31 states have different rules for self-driving cars, creating a patchwork that makes consistent safety oversight nearly impossible. Bloomberg put it plainly: “We Still Don’t Know if Robotaxis Are Safer Than Human Drivers.” That uncertainty cuts both ways. It is a reason for caution — but it is also a reason not to assume the status quo of human driving is the safe choice.