For more than a year, Tesla has shielded details about its robotaxi crashes from public view. Now, the company has published new details in a federal database about 17 incidents, which took place between July 2025 and March 2026. In at least two of them, Tesla’s human employees appear to have played a hand in the crashes by remotely driving the otherwise autonomous cars into objects on the street. In both crashes, which happened in Austin, “safety monitors” were in the vehicles’ passenger seats to oversee the still-fledgling self-driving tech, and no passengers were riding in the cars. Both crashes occurred at speeds below 10 miles per hour. The new details were first reported by TechCrunch. In one incident, which took place in July 2025, the safety monitor experienced “minor” injuries after a remote worker drove the Tesla up a curb and into a metal fence at 8 mph. The monitor, who had requested help from Tesla’s remote driving team after the car stopped on the side of a street and wouldn’t move forward, was not hospitalized, Tesla reported. The other incident, in January 2026, happened after a safety monitor requested navigation help from the remote team. The remote driver took control and drove the car straight into a temporary construction barricade at 9 mph. The crash left the robotaxi's front left fender and tire scraped up, but Tesla didn’t report any injuries. Tesla, which does not have a public relations team, did not respond to WIRED’s request for comment. The new details draw attention to an often misunderstood but safety-critical part of autonomous vehicle operations: the human backstops who remotely monitor the robot cars and intervene when they get into trouble. All US self-driving operators maintain these remote teams, according to letters submitted to a US senator earlier this year. But Tesla appears to be an outlier because it more frequently allows these remote workers to directly drive the cars. Other companies typically allow their workers t