One Week with the Cybercab: What Riders Found and What Regulators Want

On September 4, Tesla put the production Cybercab into service without a press conference, without a livestream, and without a news release. Four people, two cars, and a fifteen-minute walkthrough were enough: the first invited riders posted videos, and the rest of the world watched them. A week later, the picture is coming into focus — what it is like to ride in, where the software still hesitates, and why regulators in Washington have started asking questions.

A Launch That Whispered #

Tesla has a history of loud product reveals, from the Model 3 delivery theater to the Cybertruck’s window-smashing stunt. The Cybercab launch was the opposite: a small gathering in Austin where chief engineer Eric Early walked through the vehicle while a handful of guests filmed everything. Some watchers initially wondered whether the car had been released at all. It had — the first production units were already moving through Austin, and by the end of the first week the fleet had grown to 45 registered vehicles with more on the way.

Inside the Cabin: Two Seats, One Screen, Two Buttons #

AutoLab asked Spencer White, who previously drove a Tesla across the United States, to test the Cybercab as a passenger. His verdict on the interior is blunt: two seats, one screen, and almost nothing else. There is no steering wheel, no pedals, no mirrors, and no physical controls beyond two buttons — one below the screen for the windows, and one in the headliner with braille markings for the hazard lights. Even by Tesla’s minimalist standards, this cabin is sparse.

“Here’s my first full Cybercab ride. This vehicle is a game changer.” — Sawyer Merritt, on X, September 4

A Button That Pulls Over Instead of Stopping #

The hazard button behaves differently than in a normal car. Pressing it does not trigger an immediate brake; the Cybercab instead finds a safe spot, pulls over, switches on the hazards, and ends the trip — functionally the same as tapping “Pull Over” on the touchscreen. That is sensible in most situations and unsettling in the one case where you genuinely need the car to stop right now. Whether that trade-off is acceptable is precisely the kind of question regulators are now examining.

Finding Your Car in a Crowded Lot #

Wait times were longer than a typical ride-hail request, but locating the car was easy. The light bar on the nose glows in a color that matches the Robotaxi app, and the app shows an arrow that points the way in real time — enough to spot the vehicle across a busy parking lot. The butterfly doors then open automatically as you approach. Riders have noted that the doors do not avoid people or objects nearby, so a careless swing can clip something in a tight space.

The Software Is Cautious — Almost Too Cautious #

In a pedestrian-suddenly-steps-out test, the car reacted fast and stopped cleanly without hitting anyone. The trade-off is that the driving style is extremely careful: intersections are crossed slowly, and pedestrians get a wide, early brake. It feels safe, but it can feel hesitant, and riders expect over-the-air updates to smooth out the rhythm over time. For a service that must eventually run 24 hours a day, the balance between caution and flow is one of the things Tesla has to get right.

Glitches, Quirks, and “Things I Wasn’t Supposed to See” #

The first week also produced its share of odd moments. A rider named Matthew Skrzypczak got into a Cybercab and found the display showing a virtual joystick for manual driving — a maintenance interface, judging by the fender cameras still pointed at a Supercharger. “I have a feeling I wasn’t supposed to see this,” he wrote, before trying to drive the car himself. Another post showed a Cybercab stranded in the middle of a road, blocking traffic instead of pulling over. Neither incident was dangerous, but both were reminders that the first weeks of any robotaxi service are rarely seamless.

“I have a feeling I wasn’t supposed to see this… but I try to manually drive the vehicle anyway!” — Matthew Skrzypczak, on X, September 8

The Regulators Knock #

The bigger story of the week came from Washington. The day after the Austin launch, the National Highway Traffic Safety Administration announced an inquiry into the Cybercab’s compliance. NHTSA is reviewing Tesla’s certification process and the technical data behind it, asking whether the company’s self-certification under federal motor vehicle safety standards holds up — especially the decision to build a car with no steering wheel or pedals when those parts are still written into the rules.

A Self-Certification Under Scrutiny #

The United States relies on manufacturer self-certification: automakers declare that their vehicles meet federal standards, and regulators audit afterward. The problem for Tesla is that most of those standards were written for cars with human controls. NHTSA proposed dropping the brake-pedal requirement for autonomous vehicles back in June, but the rule has not taken effect, so the old requirement still applies on paper. Tesla argues that some outdated standards should not apply to a vehicle with no driver at all; NHTSA wants to see the reasoning behind that judgment.

Specs and Costs on Paper #

The certified numbers are impressive. Per EPA documents, the Cybercab carries a 48 kWh lithium-ion battery, a 163 kW (219 hp) front motor, and weighs 1,412 kg, with a combined range of about 673 km and energy use of 10.2 kWh per 100 km — efficiency that matters when a car runs all day. Tesla targets a price below $30,000 and operating costs around $0.20 per mile, which would make a given trip roughly half the price of Uber in the same city. The cabin also hides a Starlink V5 terminal. Tesla insists the satellite link is not part of the driving stack — all perception and decision-making happen on local AI hardware, even without any network — but it does give riders fast internet, 4K streaming, and gives the fleet remote diagnostics and map updates. Some industry voices note the extra hardware adds cost that cuts against the low-price story.

The Fleet So Far #

Operation is still limited: a geofenced area in Austin, a couple of dozen vehicles on the books, and plans to expand toward 1,000 cars under review. The service runs from 6:00 to 22:00, seats only two passengers, and excludes children under 13. Tesla AI lead Ashok Elluswamy has said 24-hour operation could follow next month once the FSD v15 software iteration is integrated. Riders who have tried it mostly praise the price — some posted bills well below what a traditional ride-hail app would charge for the same route — while complaints so far center on route-planning errors, missed destinations, and long waits.

What the First Week Tells Us #

The hardware story is easy to summarize: a purpose-built robotaxi that undercuts every competitor on cost and efficiency. The harder questions are software maturity, fleet scale, and whether Austin can be copied elsewhere. Analysts note that a cheap car only becomes a profitable business if thousands of them are on the road. Tesla, for its part, says production was redesigned to be more than five times faster than conventional manufacturing — the unboxed process assembles modules in parallel and frames the car in one step, cutting the line in half — which is the argument that 45 cars today can become a flood later.

The Cybercab reminds me of the Audi RSQ in I, Robot: a car that drives itself through the city while the occupant does something else. That film imagined 2035. Here we are in 2026, and people in Austin are already tapping a destination on a screen and rolling into traffic with nobody behind the wheel. The next stop is wherever regulators, software updates, and a growing fleet take it. For a closer look at the interior and exterior details riders are describing, the visual tour of the Cybercab walks through every feature, and the specifications guide has the full certified numbers.