Tesla's Robotaxi program recorded two crashes caused not by its autonomous driving system but by the human teleoperators deployed to rescue it, according to federal safety filings unredacted this month. Both incidents occurred in Austin, Texas, at low speeds with no passengers aboard — and both followed the same sequence: the automated driving system stalled, a safety monitor called in remote help, and the teleoperator who took over drove the vehicle into a fixed object.
The disclosure arrived when Tesla reversed course on the confidentiality of its crash data. Previously, the company had filed all incident narratives with the National Highway Traffic Safety Administration (NHTSA) under a confidential-business-information designation, keeping specific details from the public. The updated filings, released in May 2026, now include readable narrative descriptions for all 17 crashes Tesla logged between the Austin launch in July 2025 and March 2026 — the most complete public accounting of the program's safety record to date.
For the majority of those 17 incidents, Tesla's automated driving system was not at fault. A significant share were rear-endings and sideswipes caused by other drivers who struck stopped or slow-moving Robotaxis. But among the crashes caused by the Robotaxi system itself or its operators, the teleoperator incidents stand out for what they reveal about the limits of human backup.
Tesla Teleoperator Crashes: What NHTSA Filings Show
The first teleoperator crash happened in July 2025, within weeks of the Austin network's launch. Tesla's automated driving system reportedly could not move forward while stopped on a city street — unable to resolve a situation on its own. A safety monitor in the vehicle called Tesla's remote assistance team. A teleoperator assumed control, gradually accelerating and turning left, then drove the vehicle up a curb and into a metal fence at approximately 8 mph. The safety monitor sustained a minor injury requiring hospitalization — the only such injury in the entire 17-incident dataset.
A near-identical sequence unfolded in January 2026. The automated driving system was traveling straight when the onboard safety monitor again requested remote navigation support. A teleoperator took over from a stop, continued forward, and collided with a temporary construction barricade at approximately 9 mph, scraping the front-left fender and tire. No injuries were reported. No passengers were present in either incident.
In both cases, a human safety monitor was physically seated behind the wheel when the teleoperator assumed remote control — meaning the backup-to-the-backup failed.
Tesla's Teleoperator Model vs. Competitors
Tesla has told lawmakers that its teleoperators are authorized to pilot Robotaxis remotely, but only at speeds below 10 mph and only for the narrow purpose of repositioning vehicles stuck in difficult positions. "This capability enables Tesla to promptly move a vehicle that may be in a compromising position, thereby mitigating the need to wait for a first responder or Tesla field representative to manually recover the vehicle," the company stated in congressional filings earlier this year.
That model is meaningfully different from the approach used by most other commercial robotaxi operators. Companies like Waymo use remote personnel in an advisory capacity — their staff can suggest route changes or flag situations, but the autonomous system retains control of the dynamic driving task and executes any maneuver itself. Tesla's model, by contrast, allows a human operator to physically pilot the vehicle from a remote location, effectively making the teleoperator the driver.
That distinction has concerned autonomous vehicle safety researchers since before the Austin launch. Philip Koopman, a Carnegie Mellon University engineering professor with more than two decades of autonomous vehicle safety research, called Tesla's teleoperation approach "inherently unreliable technology" at the time of the rollout. "Eventually you will lose connection at exactly the worst time," Koopman told Reuters. "If they've done their homework, this won't ever happen for 10 cars. With a million cars, it's going to happen every day." Former Waymo CEO John Krafcik offered a similar assessment, calling remote driving at speed "very risky" due to latency in cellular data connections.
The NHTSA data now shows that connection loss was not the failure mode in either Austin incident — the teleoperators had control but made errors a driver sitting in the vehicle presumably would not have made. Research published in Frontiers in Neuroscience in 2024 provides neurological evidence that latency in teleoperation increases cognitive load and impairs motion planning and control, with operator performance degrading measurably even at short delays. Depth perception through a remote camera feed, combined with the physical disconnection of remote operation, creates conditions that differ substantially from in-vehicle driving.
How Does Tesla's Robotaxi Teleoperator System Work?
When a Tesla Robotaxi encounters a situation its automated driving system cannot resolve — a construction scenario, an ambiguous intersection, or an unusually complex street — an onboard safety monitor can request remote assistance from Tesla's teleoperator team. The teleoperator then assumes direct control of the vehicle, steering and accelerating it remotely via a cellular data link, at speeds the company has limited to below 10 mph. Once the vehicle has been repositioned or the immediate obstacle cleared, control returns to the automated system. The two crashes in July 2025 and January 2026 both occurred during this handoff phase, with teleoperators in direct control.
Tesla Robotaxi Safety Record: Full Picture From Austin Testing
The teleoperator failures represent two of the most striking incidents in a dataset that is more nuanced than the headline number suggests. Of the 17 reported incidents spanning July 2025 through March 2026 — all involving 2026 Model Y vehicles with the automated driving system engaged and a safety monitor present — 13 resulted in property damage only, two produced no injuries, one involved a minor injury without hospitalization, and one (the July 2025 teleoperator crash) required hospitalization.
The automated driving system itself also showed spatial awareness limitations. In September 2025, it struck a dog that darted into the roadway — the animal was unharmed. In two separate incidents, the system clipped the side mirrors of parked vehicles. In another September 2025 incident, it drove into a metal chain while making an unprotected left turn into a parking lot. Austin Police Lieutenant William White publicly confirmed there have been no major crashes and no traffic citations issued to Tesla Robotaxis operating in the city, an assessment that aligns with the federal data.
If Tesla cannot trust its automated system to navigate difficult situations and its teleoperators then crash when they intervene, the program faces a compounding reliability gap — particularly as Tesla moves toward removing safety monitors entirely from its expanding fleet.
Read more: Waymo Halts Service in Five Cities: Flood Patch Fails Weeks After Recall, Permanent Fix Missing
Tesla Robotaxi Already Expanding as Questions Mount
The timing of the unredacted filings is notable. Tesla has rapidly expanded its Robotaxi program beyond Austin. The service launched in Dallas and Houston in April 2026 — in fully unsupervised mode, without safety monitors — and Tesla has announced plans to extend to Phoenix, Miami, Orlando, Tampa, and Las Vegas in the first half of 2026. The Cybercab, a purpose-built two-seat autonomous vehicle without a steering wheel or pedals, entered production at Gigafactory Texas in February 2026. Musk acknowledged last month that safety remains the single biggest constraint on expansion speed, but the pace of city rollouts has accelerated regardless.
That expansion context matters for the teleoperator question. Both Austin crashes occurred during a period when the fleet was small, supervised, and operating under heightened scrutiny. As Tesla scales toward thousands of unsupervised Cybercabs across multiple cities, the reliability of every layer of its safety architecture — including the human fallback — becomes proportionally more consequential. Koopman's pre-launch observation that connection failures will happen every day at a million vehicles may be tested sooner than anyone anticipated.
Tesla did not respond to a request for comment.
Frequently Asked Questions
Is Tesla's Robotaxi safe?
Based on NHTSA data covering July 2025 through March 2026, Tesla's Robotaxi fleet recorded 17 incidents across all causes, with only one injury requiring hospitalization. Most crashes involved the Robotaxi being struck by other drivers rather than caused by the Tesla system itself. However, two crashes were caused by remote human teleoperators who took control of vehicles after the automated system stalled — a specific failure mode that safety researchers have flagged as a risk inherent to remote operation at scale.
What happens when a Tesla Robotaxi gets stuck?
When Tesla's automated driving system encounters a situation it cannot resolve independently, an onboard safety monitor can call Tesla's remote assistance team. A teleoperator then takes direct control of the vehicle remotely via a cellular data link, operating at speeds below 10 mph to reposition the car. This differs from competitors like Waymo, whose remote personnel advise the system rather than directly piloting it. Two Austin crashes in 2025 and 2026 occurred precisely during this teleoperator handoff phase.
Why did Tesla unredact its Robotaxi crash reports?
Tesla had previously filed all incident narratives with NHTSA under a confidential-business-information designation, keeping specific details from public view. In May 2026, the company reversed that policy and removed the confidentiality designations, making narrative descriptions of all 17 crashes publicly available for the first time. Tesla has not publicly explained the change.
What is a robotaxi teleoperator?
A robotaxi teleoperator is a remotely based human operator who can assume direct control of an autonomous vehicle during situations the automated system cannot handle. Tesla employs teleoperators who can pilot Robotaxis from a distance at speeds under 10 mph. Unlike some competitors whose remote staff provide guidance to the autonomous system, Tesla's teleoperators physically steer and accelerate the vehicle via a remote link — a model that Carnegie Mellon safety researcher Philip Koopman has described as "inherently unreliable technology" because of its dependence on cellular connectivity and the cognitive challenges of operating a vehicle at a distance.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
