Self-Driving Cars Are Logging More Crashes Than Ever — Here's What the Data Actually Tells Us
January 2026 brought a new record: 118 autonomous vehicle crashes reported to NHTSA in a single month — the highest monthly total since the federal government began requiring crash reporting in 2021. Headlines ran with the number. Social media called it proof that self-driving cars are dangerous. But like most statistics involving new technology, the story behind the data is more complicated — and more important — than the headline suggests.
Building your ideal collection? Create your own Dream Garage and design the lineup you've always wanted.
The Record That Needs Context
According to NHTSA data through April 15, 2026, fully self-driving car deaths totaled only 2 since reporting began. However, cars equipped with Level 2 advanced driver assistance systems — which are not fully automated and still require driver engagement — have been involved in 56 deaths.
Of 2,052 autonomous vehicle incidents reported to NHTSA that contain narratives, autonomous vehicles were solely at fault in only 4% of cases when accidents involved other road users. Only 14 human injuries have resulted from the 234 incidents where an AV was at fault.
Read also: Telematics Insurance Growing 18.9% Annually — AI Analyzes Driving Behavior for Dynamic Pricing
The rising crash count is largely a function of one thing: more vehicles on more roads. As the largest AV operator in the U.S. with around 2,500 vehicles on the road, Waymo unsurprisingly accounts for the highest number of incidents — 1,820 through 2025. But Waymo has also driven the most miles, with 220.6 million rider-only miles without a human driver through March 2026.
Want more car insights like this? Explore the latest stories in our Journal.
Why Comparing AV and Human Crash Rates Is Complicated
The comparison that most people want — are self-driving cars safer than human drivers? — is genuinely difficult to make cleanly. AV operators like Waymo must report any physical contact that results or allegedly results in any property damage, injury, or fatality, while most human crash data require at least enough damage for the police to file a collision report. NHTSA also estimates that 60% of property damage crashes and 32% of injury crashes involving human drivers aren't reported to police at all.
In other words: AV companies are held to a stricter reporting standard than human drivers. The data isn't apples to apples — and that gap inflates AV crash counts relative to human benchmarks.
The Distinction That Actually Matters
Not all automated vehicles are the same. The crucial line runs between Level 2 systems — driver assistance features like Tesla's Autopilot and GM's Super Cruise, which require an alert human behind the wheel at all times — and fully autonomous systems like Waymo's robotaxi, which operates without any human driver.
The line between assistance and autonomy has often been blurred, leading to misunderstandings and accidents. Tesla crashes have led to fatalities, NHTSA investigations, and significant public scrutiny. Common incidents have included crashes with stopped emergency vehicles, phantom braking, wrong-way driving, and failures to detect pedestrians and cyclists.
The distinction matters enormously for consumers making buying decisions. A vehicle sold with "autopilot" or "self-driving" language is not a vehicle that drives itself. Misunderstanding that boundary is where real accidents happen.
Are you a dealer looking to reach serious buyers? Discover what CarCupid offers For Dealers.
A January Incident That Tells the Story
In January 2026, a Waymo vehicle struck a small child in Santa Monica during morning school drop-off hours. The child ran into the roadway from behind a double-parked SUV, directly into the vehicle's path. The Waymo braked hard, slowing from 17 mph to under 6 mph before impact. The child suffered minor injuries and stood up immediately afterward. NHTSA opened an investigation to assess whether the Waymo AV exercised appropriate caution given its proximity to the school and the presence of young pedestrians.
Read also: NHTSA Updates Safety Ratings with New Driver-Assistance Tech — Tesla Model Y First to Pass 2026 Tests
The incident generated significant coverage — and illustrates exactly why the data is hard to interpret. A human driver facing the same scenario would rarely be investigated. An AV is held to a higher standard of anticipatory caution, which is arguably appropriate for a system marketed on safety.
What the cumulative data suggests is this: fully autonomous vehicles are not yet safer than the best human drivers, but they are likely already safer than average human driving — particularly in terms of impairment, distraction, and fatigue. The honest assessment of autonomous vehicle safety in 2026 is that it's getting better, the data is improving, and the comparison to human drivers is more favorable than the monthly crash count suggests.
