In 2023, 12,429 people died in alcohol-impaired-driving crashes on American roads, close to a third of all traffic deaths that year. Thousands more died because someone was distracted or too tired to react in time. These are failures of human attention, and decades of enforcement campaigns and public service advertising have only chipped away at them.
A technology now operating at commercial scale removes that failure mode from the vehicle entirely. Autonomous vehicles (AVs) have moved well past the stage of promising better roads. The results are measured and published on a rolling basis, and anyone can check them.
Waymo’s Growing Track Record of Safety
Waymo has driven 220.6 million rider-only miles with no human in the driver’s seat, and it now delivers more than 500,000 paid rides a week across 11 U.S. cities. Two years ago that figure was a tenth of its current size. More rides provide more data underscoring Waymo’s strong safety record.
Waymo’s June 2026 analysis compares the Waymo to human drivers across five of its operating geographies, including Phoenix, San Francisco, Los Angeles, Austin, and Atlanta, over the same period. The Waymo Driver was involved in 94% fewer crashes causing serious or fatal injuries, 82% fewer crashes involving any reported injury, 82% fewer airbag-deployment crashes, and 93% fewer injury crashes involving pedestrians. Every comparison holds regardless of which party was at fault.
This record rests on federal crash reporting measured against state police records, published with confidence intervals and reviewed by outside researchers.
Waymo reports every qualifying crash to NHTSA under the agency’s Standing General Order, which requires submission within 24 hours. The human benchmark comes from police-reported crash records in California, Arizona, and Texas, the same states where Waymo operates, adjusted by vehicle type and road type so the two populations are comparable. The methodology behind it has been peer-reviewed in Traffic Injury Prevention, and Waymo publishes its confidence intervals rather than only its headline percentages.
The Insurance Industry Ran the Numbers Too
Reinsurers have no ideological stake in this debate. They price risk, and they lose money when they price it wrong. So it matters that Swiss Re, one of the largest reinsurers in the world, reached the same conclusion from an entirely different dataset.
Swiss Re compared Waymo’s liability claims against human benchmarks drawn from more than 500,000 claims and over 200 billion miles of exposure. The Waymo Driver produced 88% fewer property damage claims and 92% fewer bodily injury claims than the general driving population. Against a harder benchmark of modern vehicles equipped with automated emergency braking and lane-keeping assistance, the gap barely narrowed: 86% fewer property damage claims and 90% fewer bodily injury claims.
That second comparison is the one legislators should sit with. The Waymo Driver outperforms the safest human-driven cars on the road, operated by human drivers benefiting from the best assistance technology the auto industry currently sells.
The Insurance Institute for Highway Safety came at the same question from a different angle, using crash reports instead of insurance claims, and landed in the same takeaways. IIHS researchers examined crash data that Waymo, Cruise, Zoox, and others had filed with NHTSA between 2021 and 2024, stripping out incidents where the automation wasn’t engaged, the vehicle wasn’t on a public road, or no real crash occurred. They then applied a “would a reasonable person have called the police” standard to what remained, since self-driving companies are required to log minor scrapes that human drivers routinely never report.
Comparing Waymo’s driverless mileage in Phoenix, San Francisco, Los Angeles, and Austin against human drivers in those same cities, IIHS found that Waymo vehicles were involved in 68% fewer police-reportable crashes per mile driven than human drivers, with single-vehicle crashes down 85% and injury crashes down 81%. Notably, IIHS is funded by auto insurers whose business depends on getting crash risk right, the same incentive that adds credibility to Swiss Re’s research. Two independent methodologies of Swiss Re and IIHS both arrive at the same key conclusion: autonomous driving is consistently safer than human driving.
We Should Advance Safety, Not Seek Perfection
The technology is, of course, not flawless. Recently it was reported school districts in Austin and Atlanta have documented Waymo vehicles illegally passing stopped school buses, and NHTSA opened an investigation. Waymo identified the software issue, updated its fleet, and filed a voluntary recall covering more than 3,000 vehicles. Its chief safety officer said at the time that holding the highest safety standards means recognizing when the company’s behavior should be better. Waymo has since run joint data collection with Austin ISD on school bus light patterns, and the federal review remains open as Waymo continues to work to address the issue.
When a human driver rolls through a school zone at speed, no software update follows, and no federal database logs it. The question that matters is whether these vehicles are safer than human drivers, and on that question the evidence is clear.
The current moment is a strange one. Waymo’s incredible safety record exists, it is public, and yet legislators across the country are still writing laws and rules that at least undermines the service, an in many cases, prevent it being brought to market at all.
For example, many AV opponents and skeptics argue that self-driving taxis should be required to have a paid driver for all of their rides, or set arbitrary quotas that would force self-driving taxi companies to offer human drivers. Requiring a human backup for a system proven safer than humans is like telling a company it may use computers as long as a typist sits beside every screen, ready to take over.
If a consumer chooses a Waymo specifically for the driverless experience, which many surely do, they will naturally be confused and disappointed if a vehicle with a human driver arrives instead.
Teamsters locals have backed exactly this requirement in Massachusetts, Illinois, and Washington, D.C., and Kentucky, New York, and the District still require a driver for self-driving taxis. Fundamentally, as a political force the Teamsters are doing whatever they can to obstruct AV deployment wherever and as much as they can.
Nobody making these arguments has produced data showing that a human in the seat improves outcomes. That absence is the subject of the next post in this series.
Every state that treats proven safety as a reason for more caution rather than less is making a choice about how many of those crashes it is willing to accept. Lawmakers who campaign on safer roads now have a straightforward way to deliver them, and the evidence to justify doing so.
Image via Unsplash.