NewsRoboticsDevelopingNew
Tesla publishes first Robotaxi safety report, claims no injuries over 3.1 million driverless miles
Tesla's new Robotaxi safety report says its cars drove 3.1 million miles without an onboard safety monitor with no injuries, but the figures are self-reported, include remote-assisted trips, and use a narrower injury count than its human benchmark.
Evidence: 5 sources · 1 official record · 4 independent reports · 4 open questions
| Date | Event |
|---|---|
| 22 January 2026 | Report window begins |
| Early September 2026 | NHTSA asks how the Cybercab was certified |
| 18 October 2026 | Report data cutoff |
| 21 October 2026 | Report published (Tesla says 3,112,541 miles, zero injuries; not independently audited) |
| 30 October 2026 | Tesla's reply to NHTSA due |
What happened
- On 21 October, Tesla published its first Robotaxi safety report, covering 22 January to 18 October 2026.
- Tesla says its robotaxis drove 3,112,541 miles in that period with no safety monitor inside the car, about 2.1 million of them in Austin.
- The company reports zero injuries caused by a Robotaxi, zero airbag deployments, and no pedestrians or cyclists hit.
- The report lists Las Vegas, Phoenix, Jacksonville, and San Antonio as the next cities, without launch dates.
- Separately, the US National Highway Traffic Safety Administration (NHTSA) is reviewing how Tesla certified its Cybercab, which has no steering wheel or pedals.
Why it matters
Robotaxis are the largest public deployment of AI-controlled robots on open roads, and safety data published by the operator itself is now a central part of how the public and regulators judge them.
What is confirmed and what is claimed
Confirmed
- Tesla launched a Robotaxi safety report on 21 October 2026 covering 22 January to 18 October (Tesla's report as described by Drive Tesla Canada, Not a Tesla App, and TeslaNorth; announced on X by Tesla AI executive Ashok Elluswamy). 123
- The report counts only miles with no safety monitor inside the car, and includes trips where a remote operator assisted or took over (as described by Drive Tesla Canada). 12
- Tesla has filed crash reports with NHTSA, including a low-speed Cybercab crash into a water-filled pothole in Austin with a safety monitor aboard and no passengers (NHTSA data reported by KXAN and the Austin American-Statesman). 4
- In early September NHTSA opened a query into how Tesla certified the Cybercab as meeting federal safety standards (reported by TechCrunch and Electrek). 5
Claimed, not independently verified
- The zero-injury, zero-airbag, and zero pedestrian or cyclist figures are Tesla's own and have not been independently audited. 2
- Tesla estimates that human drivers covering the same miles would have caused about 9.1 injury collisions, 4.1 airbag-deployment collisions, and 0.52 collisions with pedestrians or cyclists, based on Texas police crash data. Tesla discloses that its own injury metric counts only collisions where the Robotaxi made first contact, while the human benchmark counts all collisions. 2
- Elluswamy's announcement cited 3.3 million miles; that appears to reflect the report page's live counter, which Tesla says is estimated at the fleet's current pace, rather than the 18 October cutoff. 2
Tesla has started publishing its own safety statistics for its Robotaxi service, which now runs without a safety monitor on board in several Texas and Florida metro areas. The first report arrived on 21 October, the day Tesla was scheduled to report third-quarter results.
What the report counts
The headline figure is about 3.1 million miles driven with nobody from Tesla inside the car, from 22 January to 18 October. Over that distance Tesla reports no injuries caused by a Robotaxi and no airbag deployments. It compares this with an estimate of how many injury crashes human drivers would have had over the same miles, using Texas police data.
The method matters as much as the result. The report includes trips where a remote operator helped or took control, so these are not miles driven entirely by software. Tesla's injury count includes only crashes where the Robotaxi made first contact, while the human benchmark includes all crashes. And the data comes from Tesla, not an independent auditor.
Context
Tesla does file crash reports with NHTSA, as all US operators of automated vehicles must, and local reporters have tracked them. Those filings include a low-speed Cybercab crash into a flooded pothole in Austin. NHTSA is also asking how Tesla certified the Cybercab, which has no steering wheel or pedals, as meeting federal safety standards.
For how to read company-reported figures, see our guide to robot demos versus deployments.
What to watch
Independent analysis of the report's method, the details of Tesla's reply to NHTSA due by 30 October, and launch dates for Las Vegas, Phoenix, Jacksonville, and San Antonio.
What we don’t know yet
- How often remote operators intervened, and how those trips affect the safety comparison.
- How the report treats crashes and miles from rides with a safety monitor on board, which it excludes.
- When the four named cities will launch, and whether NHTSA's Cybercab review leads to any action. Tesla's response was due by 30 October after an extension, according to Not a Tesla App.
- Reports differ on when the Cybercab pothole crash happened, in August or September.
Evidence and sources
- Independent reporting Tesla Launches Robotaxi Safety Report, Claims Zero Injuries Over 3.1 Million Autonomous Miles
- Primary announcement Announcement of the Tesla Robotaxi safety report
- Independent reporting Tesla Names Next Robotaxi Cities, Shares Miles Driven & Safety Data
- Independent reporting NHTSA: Tesla reports its 1st Cybercab crash in September
- Independent reporting Feds launch investigation into Tesla's Cybercab deployment