Tesla Discloses 17 Robotaxi Crashes — Two Caused by Remote Operators
5 min read read
Tesla has disclosed 17 crashes involving its Cybercab robotaxi fleet to federal regulators since August 2025 — and two of those incidents happened while a remote human operator was at the wheel. New NHTSA filings reviewed by TechCrunch reveal a safety record that still lags behind human drivers by a significant margin, even as the company accelerates its autonomous fleet expansion across Texas cities.
The disclosures, which were previously partially redacted, show Tesla's Austin-based robotaxi fleet is accumulating crashes at a rate of 1 per ~57,000 miles — compared to the national human-driver benchmark of 1 per 229,000 miles, roughly 4 times worse.
Two Crashes Linked to Remote Operators
The most striking detail in the NHTSA filings is not the total crash count, but two incidents in which a teleoperator — a human remotely controlling the vehicle — was directly involved in the collision.
| Date | Location | What Happened | Speed |
|---|---|---|---|
| July 2025 | Austin, TX | Teleoperator drove vehicle up a curb and into a metal fence | Low speed |
| January 2026 | Austin, TX | Hit a construction barricade | 9 mph |
Tesla describes the teleoperator function as a way to "promptly move a vehicle that may be in a compromising position." In practice, these disclosures reveal that remote operation itself introduces a new failure mode — one that regulators and fleet operators will need to account for as the industry scales.
"This capability enables Tesla to promptly move a vehicle that may be in a compromising position."
— Tesla statement to NHTSA, on the remote teleoperator function
Fleet Availability Problems in Dallas and Austin
Beyond the crash data, newly surfaced operational metrics paint a picture of a fleet that is still far from consumer-ready at scale. In Dallas, a 36-minute wait for a robotaxi was reported, and one trip that should have taken 20 minutes ended up consuming roughly 2 hours due to routing and availability delays.
In Austin, the data is similarly sobering:
- In 27% of checks, zero available cars were found in the service zone
- More than a 15-minute wait was recorded in over 50% of ride attempts
For context, Austin currently hosts roughly 50 Tesla robotaxis — compared to 250+ Waymo vehicles operating in the same city. That supply gap alone explains much of the availability problem.
How Tesla's Crash Rate Compares
The NHTSA data enables a direct comparison to conventional driving benchmarks. The figures are based on verified filings, not Tesla's own marketing claims.
| Driver Type | Crash Rate | Context |
|---|---|---|
| Tesla Robotaxi (Austin fleet) | 1 per ~57,000 miles | Based on 17 crashes since Aug 2025 |
| Average U.S. Human Driver | 1 per ~229,000 miles | NHTSA national baseline |
| Difference | ~4x worse | Per TechCrunch analysis of NHTSA filings |
Tesla would likely argue that its fleet operates in dense urban conditions — harder than the national average — and that 17 incidents is still a small dataset. Both points are valid. But these are the numbers regulators, insurers, and municipal partners will use when deciding whether to expand autonomous vehicle permits.
The Bottom Line for Tesla's Robotaxi Program
Tesla's decision to disclose 17 crashes — after earlier redactions — signals increased regulatory transparency and a more open posture with NHTSA. That shift is worth acknowledging. But the underlying data shows a fleet that is still learning, still dependent on human backup operators who themselves have caused two crashes, and still struggling with basic availability in its two primary launch cities.
Tesla has always framed its robotaxi rollout as a data-first, iterative process. The NHTSA filings suggest that iteration has a long road ahead — both in reducing the crash rate and in scaling fleet supply to the point where a passenger can reliably hail a ride without a 36-minute wait.
Photo: Cybercab / autonomous vehicle concept / Pexels