Sterling Anderson, GM's chief product officer, has delivered the bluntest rebuke yet of Tesla's camera-only approach to self-driving system.
Anderson ran Tesla's Model X programme and then directed Autopilot during its early years in 2015 and 2016 before co-founding self-driving company Aurora.
Speaking to The Verge after GM's technology event in New York, he called vision-only driving "a weird religious stance that ignores physics," arguing cameras cannot see through dense fog or measure an object's instantaneous velocity the way radar can.
He said radar is cheap enough that removing it makes little sense, and that training cameras alongside lidar helps vision systems improve faster, a direct challenge to Tesla's argument that cameras plus fleet data are sufficient.
Tesla began stripping radar from the Model 3 and Model Y in May 2021, launching "Tesla Vision" with Autosteer capped at 75 mph and features including Smart Summon disabled at delivery; ultrasonic sensors followed in 2022.
The split now shapes GM's own roadmap: the Cadillac Escalade IQ is due to debut the company's "eyes-off" system combining lidar, radar and cameras in 2028.
Anderson also criticised marketing language like "eyes off," arguing a driver-assist system that feels capable breeds complacency, a swipe at Tesla's "Full Self-Driving (Supervised)" branding.
America's road-safety regulator upgraded its probe into FSD's performance in reduced-visibility conditions to an Engineering Analysis covering 3.2m vehicles in March.
That's one step short of a recall, after finding the system's camera-degradation detection failed to catch sun glare, fog and airborne dust before nine crashes, including a pedestrian death in November 2023.
