I watched the NHTSA spreadsheet scroll past like a live feed and felt my stomach tighten. A calm Austin street, a quiet Model Y, and then a line item that read “collision with fixed object.” You can feel how small details in a report can change a story overnight.
I’ve followed robotaxi files long enough to know what to watch for; you should too. The newest entries — which Electrek first flagged after Tesla submitted them to the National Highway Traffic Safety Administration (NHTSA) — are small incidents on the surface but big in implication.
A Model Y struck a fixed object at 17 mph — What the reports actually show
One of the five incidents reported for December 2025 and January 2026 was a collision with a fixed object at 17 miles per hour. Another involved a stopped Tesla and a bus, a third a low-speed truck collision at 4 mph, and two were backing-into fixed objects at slow speeds.
Those are the facts on the forms. What matters is how you read them: these aren’t high-speed freeway pileups, but they are repeat, systemic errors under autonomous control. I’m not arguing for panic — I’m arguing for precise scrutiny.
Five crashes in two months in Austin — How that stack compares to human drivers
Since Tesla launched the Austin robotaxi service in June 2025, the company has filed 14 crash reports with NHTSA. Based on Tesla’s disclosure that the fleet hit roughly 700,000 cumulative paid miles through November 2025 and reasonable estimates putting it above 800,000 miles by mid-January 2026, Electrek’s math arrives at roughly one crash every 57,000 miles.
How often do Tesla robotaxis crash?
Short answer: by Tesla’s filings, about once per 57,000 miles during the Austin rollout period. For context, Tesla’s own Vehicle Safety Report says the average U.S. driver experiences a minor crash about every 229,000 miles and a major collision about every 699,000 miles. Put bluntly, the robotaxi rate is near four times higher than the minor-crash benchmark Tesla cites for human drivers.
I think about that gap like a lighthouse whose bulb is dimming: it still signals, but not with the reliability you expect when people are relying on it.
Tesla redacted narratives — Why the wording matters
Each Tesla entry in the NHTSA database is heavily redacted; incident narratives are often marked “confidential business information.” That contrasts with companies such as Waymo and Zoox, whose reports are more transparent about how and why incidents happened.
Are Tesla robotaxis safer than human drivers?
Based on the metrics Tesla published and the Austin filings, the current comparison looks unfavorable to Tesla’s robotaxi fleet. Transparency becomes the multiplier: raw numbers tell one story, but redacted narratives keep you guessing about causation, system state, and corrective measures.
One notable correction: a July report that was initially filed as “property damage only” was later updated to “Minor w/ Hospitalization,” indicating someone did get medical care after the event. That kind of revision matters when you’re weighing risk.
NHTSA probes and industry context — What regulators are watching
The NHTSA has active scrutiny across the industry. The agency opened an investigation into a January 2026 incident where a Waymo vehicle struck a child near a Santa Monica elementary school during drop-off, and it’s separately probing reports that Waymo cars failed to stop for school buses. Those probes show the regulator is treating robotaxi safety as a systemic issue, not a single-company problem.
What is the NHTSA investigating about robotaxis?
Beyond the Austin filings, NHTSA’s Office of Defects Investigation is asking whether companies exercised appropriate caution around schools and other vulnerable road users. If you follow enforcement, you’ll see probes focused as much on decision-making and edge cases as on vehicle hardware.
Here’s what I want you to hold onto: the numbers are what they are, but context and transparency change how persuasive those numbers feel. Tesla’s Austin fleet has driven a lot of miles and logged 14 official crashes since June 2025; that’s a fact. Whether that translates to an unacceptable safety gap depends on how the company explains each incident, what fixes follow, and how regulators respond.
Tesla, Electrek, Gizmodo, NHTSA, Waymo, Zoox, and even the Model Y and FSD branding are all now part of the same conversation about machine behavior on public streets. One way to think about the rollout is like a card dealer quietly reshuffling the odds while the table fills — some hands will reveal problems faster than others.
I’ll keep watching the filings and the probes so you don’t have to read every spreadsheet, but I’ll also push you to ask the hard questions when a fleet’s crash rate outpaces human benchmarks by a factor of four: what fixes are being tested, who is accountable, and how quickly will those changes reach passengers and pedestrians?
Is it acceptable for a robotaxi program to run a public test at scale when its reported crash rate is multiple times higher than an average human driver’s?