A software glitch halted a Tesla Cybercab robotaxi ride in Austin, Texas, leaving a passenger stranded without visual feedback on the vehicle’s navigation logic. The incident occurred during a weekend test drive just days after the car’s public launch. While the first trip proceeded smoothly, the second attempt revealed critical software instability that forced a manual exit and a ride cancellation. This event highlights the ongoing challenges of deploying fully autonomous vehicles in live traffic conditions, where software reliability is just as important as hardware design.
The Smooth Start and the Glitch
The initial experience with the Cybercab was largely positive. The vehicle arrived exactly on the predicted 17-minute mark, pulling up approximately 150 feet from the hotel entrance. A purple light bar illuminated above the windshield, matching the app’s interface to signal the correct vehicle. The scissor doors opened automatically, slowing down if the passenger approached too closely to prevent injury.
Once the seatbelt was fastened, the doors closed, and the ride began with a smooth, quiet cabin experience similar to a Model 3, though the rear suspension transmitted larger road bumps. Interior features included USB-C ports and window switches, though the windows only rolled down a few inches for safety. The center armrest could adjust to allow easy exit. The dashboard displayed the route and decision-making process on a 3D rendering of the car, a design choice that deemphasized the traditional instrument cluster. However, the absence of a turn signal clicking sound was noted as slightly eerie.
Read Also: Audi A8L Leads 2004 Luxury Sedan Showdown
The drop-off at a coffee shop was straightforward, with camera angles popping up on the screen to assist in exiting safely. The ride cost $12.68 for 0.8 miles, a rate of $15.85 per mile, significantly higher than typical ride-hailing services due to dynamic pricing during the debut weekend.
When the Screen Fails
The return trip to the hotel presented a different scenario. The second Cybercab appeared in the same location, and the passenger-side door opened automatically. Upon entering, the main display showed an “Aw, Snap!” error screen, reminiscent of a Google Chrome browser crash, with a warning that read, “Issue identified—resolution in progress.” Attempts to interact with the touchscreen to call support or adjust settings failed, though the physical window controls still functioned. The physical door release allowed the passenger to open the door manually, similar to a mechanism found in a McLaren.
Desperate to catch a flight, the passenger initiated the ride from the phone app. The car began driving, but the screen remained frozen on the error message. This lack of visual feedback created a sense of unease, as the passenger could not see the car’s logic, route, or intended maneuvers. For several minutes, the vehicle appeared to be returning to the hotel based on the first ride’s path, but it was actually just circling the block.
The Cybercab eventually returned to the exact pickup spot without indicating the ride was over. The app still showed the ride as waiting for entry. The passenger had to physically open the door with the handle to exit, worried the car might drive away. Once outside, the ride was terminated at no cost, and an error report was filed. The vehicle sat briefly before driving off, and the passenger took a Lyft to finish the trip, which cost $10.43 for 0.86 miles.
Read Also: Italian police receive new Lamborghini Temerario
Software-defined vehicles offer significant benefits in terms of updates and functionality, but faulty code remains a critical vulnerability. When the primary interface fails, the passenger is left without a window into the vehicle’s operational status. This disconnect between the physical vehicle and the digital interface can erode trust, even if the autonomous driving system itself is functioning correctly. The reliance on a single display for both navigation and status creates a single point of failure that, if compromised, leaves the user in a state of uncertainty about the car’s immediate actions and destination.
Comparing the Cybercab to Waymo
To provide context, a comparison was made with Waymo’s robotaxi fleet in Los Angeles. A recent Waymo trip covered 12.24 miles from the West Side to Korea Town in 29 minutes, costing $21.46, or $1.75 per mile. Unlike the Cybercab, which cannot yet drive on freeways, Waymo’s Jaguar I-Pace robotaxis are certified for interstate use. During the test, speeds reached 65 mph, and the ride felt safe throughout. The Jaguar rides rougher than the Cybercab but absorbs larger impacts better, with snappier acceleration and turning.
Waymo’s in-car experience is less automated, with screens that feel slightly dated, though the route mapping and exterior renderings are high quality. Passengers have access to entertainment apps through their accounts, similar to the Tesla setup. The key difference is reliability and scope. Tesla’s Cybercab service is still in its infancy, having opened to the public on September 4, 2026. According to the Texas Department of Motor Vehicles, there are currently 45 Cybercabs operating in the Austin area. Tesla states that pricing should decrease as more vehicles enter the fleet, but the software issues observed in the test drive suggest that technical stability is a more immediate hurdle than pricing.