Robotaxi safety FAQ
The questions everyone asks about Waymo, Tesla Robotaxi, and the numbers on the board — answered with sources, not vibes.
Has a robotaxi ever killed anyone?
No death has ever been attributed to the driving of a Waymo or Tesla robotaxi. Waymo vehicles have been involved in two fatal crashes, but investigators attributed both to human drivers — a speeding driver striking stopped traffic in San Francisco (January 2025) and a motorcyclist rear-ending a slowing Waymo in Tempe (September 2025). For perspective, human drivers kill about 107 people on US roads every day.
Is Waymo really safer than a human driver?
The best independent evidence says yes for the metrics studied so far. A July 2026 IIHS matched study found Waymo had 1.28 police-reportable crashes per million miles versus 4.06 for human drivers in the same areas and years — 68% fewer crashes and 81% fewer injury crashes. Waymo's own published analysis through March 2026 reports similar reductions across 220+ million driverless miles. These are matched comparisons, not raw counts.
Why is Tesla's Robotaxi crash rate shown as "n/a"?
A crash rate needs two numbers: crashes and miles driven. Tesla reports crashes to NHTSA as required (22 through mid-2026) but has not published its robotaxi mileage, so no honest per-mile rate can be computed. An early independent estimate from February 2026 put the rate around 9–18 crashes per million miles in Austin, roughly 4× a human benchmark, but it relied on estimated mileage and mostly monitored driving — we treat it as historical context, not a current rate.
Why does Waymo have so many more reported crashes than Tesla?
Exposure. Waymo has driven over 220 million driverless miles with thousands of vehicles since 2020; Tesla's Robotaxi service launched in June 2025 with a few dozen vehicles. Raw crash counts without mileage are not safety rates — that's why the board shows both the counts and the per-mile comparisons, and labels what can't be computed.
Do robotaxis really have to report every minor bump?
Yes. Under NHTSA's Standing General Order, automated-driving companies must file a report for qualifying contacts, including minor fender-benders and crashes caused entirely by the other driver. Human crash statistics only capture police-reported crashes — most minor human crashes never enter any database. Comparing raw robotaxi reports to police-reported human rates therefore overstates the robotaxi numbers; matched studies correct for this.
Does NHTSA data say who was at fault?
No. NHTSA's crash reports record involvement, not fault — a robotaxi rear-ended while stopped at a red light still generates a report. Press tallies of raw NHTSA totals include many crashes caused by human drivers hitting a robotaxi. Fault determinations, where they exist, come from police reports, insurers, and investigators, which is why we link a source beside every number.
Are consumer Teslas on Autopilot or FSD counted?
No. Privately owned Teslas running Autopilot or "Full Self-Driving (Supervised)" have a licensed human driver legally responsible at the wheel — a different product with its own separate crash record. This site tracks driverless taxi services only. News coverage frequently conflates the two; the scoreboard keeps them separate.
What counts as a "city in service"?
A metro area where the general public can hail a fully driverless ride today, with no human safety driver in the car. Announced-but-unlaunched cities are listed separately on the map. Tesla's SF Bay Area service is excluded because rides there still require a human safety driver under California permits.
Who runs RoboTaxi Report? Is it affiliated with Waymo or Tesla?
RoboTaxi Report is independent and not affiliated with, sponsored by, or endorsed by Waymo, Tesla, or any autonomous-vehicle company. The board applies the same methodology to both companies, links a source beside every number, and corrects errors when readers send evidence. Full definitions live on the About the numbers page.
How often is the data updated, and can I use it?
The scoreboard refreshes weekly from NHTSA crash reports, company announcements, and press reporting. The underlying data is free to download as JSON, and media and researchers may cite or republish the numbers with attribution to RoboTaxi Report and a link to robotaxireport.com — see the media page for details.