Tesla filed one new “Robotaxi” crash report with NHTSA in its February submission, bringing the total to 15 incidents since the program launched in Austin in June 2025. That’s a significant drop from the five crashes reported in the last report, but without specific mileage data from Tesla, there’s no way to determine whether the fleet is actually getting safer.
The new incident involved a Model Y hitting a fixed object at 9 mph in January 2026 with the autonomous driving system engaged. No injuries were reported.
One crash vs. five — but context matters
Last month, we reported that Tesla had filed five new crash reports in a single monthly submission, covering incidents from December 2025 and January 2026. Those five included collisions with a bus, a heavy truck, a pole, and two fixed objects — all at low speeds with the autonomous system verified engaged.
As we previously stated, this month’s crash involved a Model Y hitting a fixed object at 9 mph in January 2026 (this month’s report covers the second half of January and first half of February) with the autonomous driving system engaged. No injuries were reported.
This month’s single report looks better on the surface. But the number of crashes in a given month is meaningless without knowing how many miles the fleet drove during that period.
Tesla’s “Robotaxi” service was also shut down for several days in January due to an ice storm that hit Austin, further complicating any attempt to compare month-over-month performance. Fewer miles driven naturally means fewer opportunities to crash. There is no way to know for sure that mileage increased — or even maintained — from one month to the next.
Tesla does not regularly disclose specific mileage figures for its robotaxi fleet. The company disclosed approximately 500,000 cumulative miles through November 2025 in a one-time filing, and we extrapolated approximately 800,000 miles by mid-January when analyzing the previous batch of crashes. At that rate, the fleet’s crash rate was roughly one incident every 57,000 miles — compared to one crash every 500,000 miles for the average American driver, according to NHTSA data.
9 months in: ~35 vehicles, low availability, full supervision
The lack of mileage data is especially frustrating when put in the context of how small Tesla’s robotaxi operation remains.
Based on the latest crowdsourced data from independent trackers who reverse-engineered Tesla’s ride-hailing app, the company is operating approximately 35 vehicles in Austin, nine months after launch. Availability remains low — our status check last month found it below 20%.
Every single one of those vehicles operates with an in-car safety supervisor who has a finger on a kill switch. That supervisor is presumably preventing additional crashes beyond what we’re seeing in the NHTSA data — making the actual per-mile crash rate of the unsupervised autonomous system even worse than the reported numbers suggest.
Tesla is also operating a single unsupervised robotaxi that is limited to a small section of the service area in Austin. Musk announced in January that Tesla had started offering rides without a safety monitor, but the “unsupervised” fleet is limited to a single vehicle operating at a time in a very limited part of the Austin service area.
For comparison, Waymo operates over 2,500 fully driverless robotaxis across multiple US cities with no safety monitor in any vehicle. The company has logged over 127 million autonomous miles, and independent research published in peer-reviewed journals shows Waymo’s crash rate is 85% lower than human drivers for injury-causing incidents.
Tesla continues to redact all crash narratives
As with every previous filing, Tesla redacted the entire narrative section of its new crash report, marking it as “confidential business information.” Every other autonomous vehicle operator filing with NHTSA, including Waymo, Zoox, and Aurora, provides full descriptions of what happened in their incidents.
This matters because crash narratives contain the most valuable information for understanding what went wrong: the circumstances leading up to the crash, what the autonomous system did or failed to do, and whether a safety supervisor intervened. Without narratives, all we know is that a Model Y hit a fixed object at 9 mph. That tells us almost nothing about the system’s actual capability.
Electrek’s Take
One crash report in a month versus five is a better headline for Tesla, and we hope it reflects genuine improvement in the system. But hope is not data.
The fundamental problem remains unchanged: Tesla does not disclose enough information for anyone, regulators, researchers, or the public, to evaluate whether its “Robotaxi” service is getting safer. We don’t know the monthly mileage. We don’t know how many interventions the safety supervisors are making. And we can’t read any of the crash narratives because Tesla redacts them all.
What we do know is this: nine months after launch, Tesla is still running about 35 vehicles with low availability, all supervised, in a single city. The single unsupervised vehicle is confined to a tiny portion of Austin’s service area. The crash rate we can calculate, one every 57,000 miles, is still roughly nine times worse than the average human driver, and that’s with a safety monitor presumably preventing worse outcomes.
If you focus only on fender-benders, as most of the accidents appear to be, you are about 4x times higher than the human crash rate, but again, that’s with a human supervisor inside the vehicle.
Tesla is nowhere close to ready to expand without supervision. The company needs to either start disclosing comprehensive mileage and intervention data, or accept that the public will continue to judge the program by the only data available — and right now, that data does not paint a flattering picture.
FTC: We use income earning auto affiliate links. More.
Comments