GM’s product chief, who once ran Tesla Autopilot, calls camera-only self-driving ‘a weird religious stance’

Sterling Anderson, General Motors’ chief product officer, has called dropping radar in favor of camera-only self-driving “a weird religious stance that ignores physics.” Anderson ran Tesla’s Autopilot program in 2015 and 2016, after leading the Model X program, so he is criticizing the approach from inside its history. The remark came in an interview published by The Verge on October 27, after GM’s technology event in New York.

Tech news in one quick read: Get the most interesting stories in startups, tech and programming, each with a short TLDR, in one daily email from TLDR.

Subscribe to TLDR →

We may earn a commission if you sign up.

Electrek’s Fred Lambert reported the comments on October 29 and sided with Anderson. The argument lands on a live regulatory question about how well camera-only systems cope when they cannot see, and on a business question for GM about which sensors its own driver-assist products will carry into the next decade.

Anderson’s case against vision-only

The speaker is a rare critic with a résumé on both sides of the argument. Anderson later co-founded the self-driving company Aurora, and GM hired him as chief product officer in 2025. Electrek notes that Bloomberg reported last year that GM chief executive Mary Barra sees him as a frontrunner to become GM’s next CEO. The interview itself is on The Verge, which Electrek cites; that page could not be opened for this report, so the quotes below come from Electrek’s account of it.

According to Electrek’s account, Anderson said a system that scales back to vision alone invites trouble in certain situations, because physics works against it. He argued that radar is cheap, which makes removing it hard to justify, and that cameras cannot see through dense fog or measure velocity the way radar does. He added that lidar helps camera systems improve faster when the two are trained together, which pushes back on Tesla’s position that cameras plus fleet data are enough.

Anderson also steers clear of eyes-off marketing, arguing that a driver-assist system that feels capable encourages complacency in the person behind the wheel. GM’s plan for its own eyes-off system is the Cadillac Escalade IQ in 2028, which is meant to combine lidar, radar and cameras.

Electrek is not a neutral reader here. Lambert writes that Anderson is right, and ties Tesla’s continued defense of vision-only to Musk’s 2019 description of lidar. Anderson’s own claims are about physics and engineering trade-offs, and he did not, in the account Electrek gives, say that Tesla’s current cars are unsafe or name a specific product. Tesla has not publicly answered the remarks in any source read for this report.

How Tesla ended up without radar

Tesla started removing radar from the Model 3 and Model Y in May 2021. The Tesla Vision launch came with limits: Autosteer capped at 75 mph, a longer minimum following distance, and Smart Summon and Emergency Lane Departure Avoidance disabled at delivery. Ultrasonic sensors followed in 2022.

Former employees later said engineers had tried to persuade Elon Musk not to drop radar, and that he overruled them. Musk told Electrek at the time that a very high-resolution radar does not exist, a claim Electrek disputes by pointing to many such units now on the market. Electrek also recalls that Musk called lidar a fool’s errand in 2019, even as Tesla now builds its own fleet maps after years of mocking maps.

NHTSA’s reduced-visibility probe of Full Self-Driving

The physics argument has a regulatory counterpart. NHTSA opened Engineering Analysis EA26002 on March 18, 2026, according to the agency’s opening resolution, covering an estimated 3,203,754 Tesla vehicles equipped with FSD. The Office of Defects Investigation listed nine incidents, all crashes or fires, including one fatality on November 28, 2023, and said it is concerned that the system fails to detect degraded visibility from glare and airborne obscurants, or fails to warn the driver in time.

NHTSA’s finding, as Electrek summarizes it, is that the system’s camera-degradation detection did not detect common roadway conditions that impaired camera visibility, including sun glare, fog and airborne dust, and that drivers were alerted only right before the crashes. The investigation also reaches beyond the nine original incidents: the Office of Defects Investigation said it will analyze six more recent, potentially related events, and that limits in Tesla’s data and labeling may have caused underreporting of relevant crashes.

In the reviewed crashes, the agency found that FSD lost track of or never detected a lead vehicle in its path. Tesla told NHTSA that its June 2024 update may have affected three of the nine incidents had it been installed, but the agency does not yet know when that update was deployed or on which vehicles. An Engineering Analysis is the step before a possible recall request.

Electrek sets the stakes against Waymo, which uses lidar, radar and cameras and has logged about 271 million rider-only miles, versus roughly 3 million driverless miles for Tesla since January, when it began removing safety monitors. Tesla has not publicly responded to Anderson, and whether the probe ends in a recall is undecided.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

Tech news in one quick read: Subscribe to TLDR →

We may earn a commission if you sign up.