Tesla Robotaxi Crashes: 2 of 17 Driven by Humans Remotely (May 2026)
A Tesla Robotaxi is the company's driverless ride-hailing vehicle that launched in Austin, Texas, in June 2025. Elon Musk has spent five straight years promising "no human needed" behind the wheel. This week, newly unredacted crash reports submitted to NHTSA exposed a different reality: two of seventeen disclosed Tesla Robotaxi crashes happened while a human teleoperator was remotely driving the car.
What the Unredacted NHTSA Reports Reveal
Tesla, like all autonomous vehicle operators in the U.S., is required to submit crash details to NHTSA (the National Highway Traffic Safety Administration). For months, Tesla redacted every single Tesla Robotaxi crash description as "confidential business information." On May 15, the company reversed course and unredacted all 17 crash narratives. Two of them now reveal the same pattern: the autonomous driving system (ADS) got stuck, a remote teleoperator took control from a Tesla operations center, and then crashed the vehicle.
The Two Crashes In Detail
The first incident occurred in July 2025, shortly after the Tesla Robotaxi service launched. The ADS froze on a street in Austin. The in-vehicle safety monitor called for help, a teleoperator (remote driver) took over, and the car drove up onto the curb and into a metal fence. The second incident, in January 2026, played out the same way: ADS stopped, teleoperator assumed control, and the car hit a temporary construction barricade at approximately 9 mph, scraping the front-left fender and tire.
Tesla has already admitted to U.S. lawmakers that remote operators can move vehicles below 10 mph "to move a vehicle out of a compromising position." That admission, plus the crash data, contradicts the "fully autonomous" framing that has powered Tesla's stock narrative for years.
Why This Matters for the Whole Industry
The Tesla Robotaxi disclosure landed in the same week Waymo recalled 3,791 vehicles over a software bug that allowed cars to drive into flooded roads, after one Waymo in San Antonio was swept away in a creek on April 20. Both stories point to the same uncomfortable truth: every commercially deployed self-driving service today still leans on humans (in remote rooms, in the driver's seat, or on standby) when the AI cannot decide what to do. Musk himself recently called safety "the biggest limiting factor" on scaling Robotaxi.
If you are considering a robotaxi ride, an autonomous vehicle stock, or a self-driving Tesla, the next question worth asking is which parts of the trip are actually autonomous and which are still being filled in by a human you cannot see.
Key Takeaways
① NHTSA 17 crashes unredacted - Tesla reversed its "confidential" stance on May 15 and disclosed all crash narratives.
② 2 crashes driven by humans remotely - In both, the ADS stalled, a teleoperator took over, and the car hit a fence or barricade.
③ "Autonomous" is marketing, "assisted" is reality - Musk admits safety is the biggest bottleneck to scaling.
Self-driving has never been "almost here" for as long as it has been "almost here." The gap between the marketing label and the operational reality is now visible in plain English, in the NHTSA filings themselves.
👉 Figure AI Humanoid Runs 50 Hours Nonstop, Zero Human Touch - also worth a read.
📌 Sources: TechCrunch, Bloomberg, NHTSA (2026)
댓글
댓글 쓰기