Austin runs dark at night
Over the past two weeks, the Robotaxi Tracker — a community-driven database created by engineering student Ethan McKanna, who reverse-engineered Tesla's app to monitor the robotaxi fleet — recorded 170 rides in Austin, Texas, and all of them took place with absolutely no human supervisor in the front passenger seat. The milestone marks a qualitative shift: eight months after Elon Musk announced that vehicles were operating without monitoring, the operation has now fully eliminated human presence from the front seats of the Model Y fleet.
The latest development arrives at a strategically relevant moment. Tesla has been testing the production version of the Cybercab — the vehicle without a steering wheel or pedals, purpose-built for autonomous transport — in Austin since June 2026, with safety monitors in the passenger seat. During the same period, the National Highway Traffic Safety Administration (NHTSA) proposed eliminating the mandatory pedal requirement for vehicles designed exclusively for autonomous driving, paving the regulatory path for the Cybercab.
Reverse engineering as an audit tool
The Robotaxi Tracker was born from an improvised but surprisingly effective approach. McKanna, a Texas A&M engineering student, dismantled Tesla's app communication protocol to track how many vehicles were online and operating at any given moment. What began as curiosity has become the only independent public source of data on the actual scale of the operation.
By December 2025, the tracker revealed that the fleet consisted of just 32 Model Y vehicles — far below the 500 Musk had promised by year-end. Even so, the operation had been gradually scaling up, and the latest data — 170 rides over two weeks, all unsupervised — suggests that Tesla has significantly expanded its Austin pilot fleet.
Safety data: an unsettling portrait
The accident figures reported to NHTSA paint a picture that concerns safety experts. Since the service launched in June 2025, Tesla Robotaxis have been involved in at least eight reported collisions — all with a safety monitor on board, someone whose job was to intervene in emergencies. The crash rate, calculated from the 250,000 miles Tesla itself disclosed, equates to roughly one accident every 60,000 miles.
For context, the average human driver crashes about once every 500,000 miles. This means Tesla's autonomous fleet, even with a trained supervisor in the passenger seat, has a collision rate more than eight times higher than human drivers. Ironically, Tesla itself uses NHTSA mechanisms to obscure crash reports, systematically redacting the narrative section of every incident report — making it impossible for the public to know who was at fault or what actually happened.
The Waymo comparison
While Tesla takes its first steps in fully unsupervised operations in Austin, competitor Waymo — an Alphabet subsidiary — operates robotaxis in six U.S. cities (Austin, Phoenix, San Francisco, Los Angeles, Atlanta, and Miami) and processes around 450,000 autonomous rides per week. Since it began operations, Waymo has accumulated more than 125 million fully driverless miles, with a crash rate of approximately one collision every 98,600 miles. Unlike Tesla, Waymo publishes substantial detail about every incident.
The difference in approach is revealing. Waymo employs a complex sensor suite — including lidar, radar, and cameras — on adapted Jaguar and Zeekr vehicles. Tesla bets exclusively on cameras, arguing that pure computer vision is sufficient for autonomous driving. The results so far suggest that Tesla's path is more ambitious in technological ambition but considerably less mature in operational safety.
What to watch
The transition from supervised to fully unsupervised operations in Austin puts Tesla before a dual challenge: validating that its pure-vision algorithm works without human intervention in a real traffic environment, and convincing skeptical regulators that safety hasn't been sacrificed for a media announcement. With the Cybercab already in production and NHTSA evaluating the removal of the pedal requirement, the next chapter could be scaled autonomous operation — or a recall that delays everything again. The question is whether Tesla's technological ambition can keep pace with the real complexity of autonomous driving.
Sources: The Verge, Electrek, Not an FSD Tracker
✓ Independent sources cross-checked and verified before publishing
