Austin, Texas — The first paying passengers climbed into Tesla’s robotaxis two weeks ago. By the third day, some of them were watching the car drift into oncoming lanes. Others felt the vehicle slam its brakes for no visible reason. A few were dropped off in the middle of intersections.
The launch of Tesla’s robotaxi service on June 22, 2025, was supposed to be a victory lap for Elon Musk’s long-promised autonomous future. Instead, it has become a test of whether the technology is ready for real roads — and whether regulators will let the company keep moving before it is.
Every robotaxi operating in Austin carries a human safety monitor in the front passenger seat. That monitor is there because Tesla knows the system is not yet fully reliable. The monitor is there because the company’s own Full Self-Driving software has a documented history of erratic behavior. And the monitor is there because federal regulators are already paying close attention.
The National Highway Traffic Safety Administration has not announced a formal investigation. But the agency has a pattern. It opens probes after incidents pile up. Incidents are piling up. Early riders have reported wrong-way driving, phantom braking, and drop-offs in active traffic lanes. Each one of those is a data point. NHTSA collects data points.
Musk has described the robotaxi network as a central piece of Tesla’s long-term strategy. Owners would add their personal vehicles to a shared autonomous fleet. The cars would earn money while their owners slept. That vision is now live in one city, with one small fleet, and with human backup drivers required in every car.
The gap between the vision and the reality is wide. The vision: millions of vehicles, no drivers, seamless trips. The reality: a handful of cars, a safety monitor in every passenger seat, and a growing list of complaints from the first wave of riders.
That gap matters for Tesla’s stock price, which has long carried a premium for future autonomy. It matters for the company’s regulatory standing. And it matters for the broader autonomous vehicle industry, which has watched Tesla promise full self-driving for years while competitors like Waymo operate limited commercial services with more conservative technology stacks.
Tesla says the service will eventually work with all current Tesla vehicles. For now, the company is focusing on a narrow rollout. It has not said how many cars are in the Austin fleet. It has not said when the safety monitors will be removed. It has not said what criteria will trigger an expansion to other cities.
What is clear is that the robotaxi launch has already drawn the kind of scrutiny that slows down autonomous vehicle programs. NHTSA has a history of issuing subpoenas, demanding data, and forcing recalls when it sees patterns of dangerous behavior. The agency does not move fast. But it does move.
For the riders who have already used the service, the experience has been mixed. Some have posted videos of uneventful trips. Others have posted videos of near-misses. The uneven quality of the rides suggests the software is not yet handling all edge cases reliably — and edge cases are what kill autonomous vehicle programs.
The next few weeks will tell a lot. If NHTSA opens a formal investigation, Tesla will be forced to hand over logs, software versions, and internal safety assessments. If the incidents continue, the company may have to pull the safety monitors out of the passenger seat and put them behind the wheel. If the incidents stop, Tesla can point to the data and argue the system is improving.
Right now, the robotaxi service is a proof of concept with training wheels. The question is whether those training wheels come off before the regulators step in.
