Tesla Inc.‘s (NASDAQ:TSLA) former data labelers and engineers do not trust the EV giant’s self-driving system as well as its safety data, according to a Reuters investigation released on Thursday.
Doubts Over System’s Safety
The news agency conducted interviews with over nine former data labelers, as well as a former Tesla engineer, who pointed to the company’s system struggling with things like avoiding emergency vehicles, as well as school buses and giving motorcycles space.
The interviewees said that they worked long hours to train the system on specific conditions to make it appear more capable than it was.
The report also pointed to multiple experts being asked by the news agency to review Tesla’s crash data, as well as compared the data with rival Alphabet Inc. (NASDAQ:GOOGL) (NASDAQ:GOOG) Waymo‘s methodology. Experts found the data to be invalid, with some pointing to the data amounting to misleading marketing.
Tesla compared its Full Self-Driving (FSD) crashes with federal crash data on the average U.S. vehicle, which are far older than Teslas, the report said. “It’s like saying: ‘My jet airplane is faster than your World War II bomber,” Phil Koopman, who is a Carnegie Mellon University engineering professor cited in the report, said.
Over seven former Tesla employees interviewed in the report said they wouldn’t trust the FSD system to drive them, the report said. There were also instances of Tesla vehicles striking animals at speed and “Mad Max” mode driving 20-30 mph over the speed limit.
Limited Access To Videos, 10x Safety Claim Flaw
The labelers pointed to incidents of Tesla vehicles nearly striking workers in a construction zone, as well as pointing to a “trauma team” that reviews incidents of near-misses with pedestrians, the report said. The interviewees recalled incidents where the vehicles nearly hit children in 2025.
Access to the videos is limited, with labelers only seeing videos they’re assigned to see, which may point towards more serious incidents that weren’t reviewed by the interviewees, the report said.
The report then pointed to a methodology employed by the automaker to compare Tesla’s airbag crashes with Federal data on crashes that required tow truck removals, which aren’t serious enough, according to the report. Tesla claims FSD is 10x safer than humans based on this data, the report said.
Tesla also counts wrecks that happen either with FSD switched on or within five seconds of the feature being turned off, in contrast with Federal requirements that mandate automakers report crashes that occur within 30 seconds of the driver-assistance system being deactivated.
Four sources cited in the report said that it could take years to scale Tesla’s Robotaxi operations, despite Musk’s claims of scaling up operations. Tesla recently rolled out Robotaxis in Houston and Dallas, Texas.
Series Of Lawsuits
The automaker faces close to 21 active lawsuits, comprising wrongful death lawsuits, as well as customers suing the EV giant over its self-driving and driver assistance systems. The exposure estimates from these legal troubles could range from $2.1 billion to $14.5 billion.
Estimates point to Autopilot/FSD crash lawsuits commanding exposures of $1-5 billion, while class action lawsuits relating to Full Self-Driving (FSD) false advertising come in at around $100-500 million.
Meanwhile, Tesla’s FSD technology also faces scrutiny from the National Highway Traffic Safety Administration (NHTSA), which moved its investigation to Engineering Analysis, a stage that usually precedes a recall, according to laws.
Dan O’Dowd, the founder of The Dawn Project, has been a vocal critic of the automaker’s self-driving pursuits, saying that the company’s vehicles with the Hardware 3 (HW3) chip were considered “legally blind” in California, following his own investigation.
Check out more of Benzinga’s Future Of Mobility coverage by following this link.
Photo courtesy: Shutterstock
Recent Comments