Tesla discloses two robotaxi crashes involving teleoperators in newly unredacted reports
Newly unredacted crash reports filed by Tesla with the US National Highway Traffic Safety Administration (NHTSA) reveal that teleoperators, remote human operators, also played a role in two incidents in the company's robotaxi fleet tested in Austin. The reports are seen as the first public glimpse of the problems Tesla has encountered in scaling up the robotaxi fleet.
The first crash took place on 14 March 2026 at a downtown Austin intersection. According to the report the robotaxi was slow to interpret a red traffic signal; the vehicle entered the intersection before the teleoperator intervened and rear-ended a sedan ahead of it. No injuries were reported, but Tesla said the communications delay in that incident was on the order of a tenth of a second.
The second crash occurred on 28 March 2026 in south Austin. During a parking manoeuvre on a narrow street, the robotaxi struck a parked pickup truck after a teleoperator issued an external reverse command. Tesla described the event as 'an incorrect synchronisation between two AI decision nodes and human intervention'. No injuries were reported.
In both reported incidents the material damage was limited. NHTSA, however, drew attention to what it called a 'transparency boundary' in the agency's audit of the Tesla robotaxi system, specifically around how teleoperator interventions are documented and reported. An NHTSA spokesperson said that the 'public-facing version of these reports contains a meaningful gap compared to information Tesla shares with the industry.'
Tesla's robotaxi fleet was launched as a pilot in Austin in June 2025. The fleet currently consists of 22 vehicles; Tesla aims to expand the Austin fleet to 100 vehicles by the end of 2026, and to launch a pilot in Phoenix in 2027. The number of teleoperators is not public, but the company confirmed that each vehicle has 'rolling-shift remote oversight access.'
The teleoperator debate has been a recurring theme in the autonomous vehicle sector over the past few years. Waymo and Cruise have acknowledged the remote-intervention role in their fleets but kept operational details private. Tesla's crash reports clarifying the direct role of teleoperators are being read as 'a redefinition of disclosure standards for the industry.'
Dr Phil Koopman, who researches autonomous vehicles at Carnegie Mellon University, told TechCrunch: 'The two events Tesla has disclosed reflect the highest level of complexity that has emerged in usage data obtained globally in robotaxi pilots.' Koopman said it was inevitable that remote-supervised operation of robotaxi systems would feed back into structural safety standards.
Alongside the US regulatory discussion, autonomous-vehicle rule-making in Europe is also entering a new phase. The European Commission has announced that it is working on a draft regulation setting out definitions of the teleoperator role within the autonomous-vehicle type-approval process due to begin in 2027. The information Tesla has disclosed is being treated as one of the core reference documents for that process.
A Tesla investor-relations spokesperson said that the unredaction of the reports was 'an expression of our commitment to transparency standards.' The company did not share teleoperator intervention rates, but stated that 'fewer than ten manual interventions are needed per 100,000 robotaxi kilometres.' That figure is below the threshold the industry accepts for 'Level 4 readiness.'
NHTSA said it would complete a technical audit of Tesla's robotaxi reporting structure within the next three months. The output report is seen as a binding document that could affect Tesla's plans for fleet expansion. Tesla said at Friday's analyst call that it would assess its end-of-2026 target 'in light of the audit outcome as well.'