Tesla is scaling back its boldest claims on autonomous driving as Elon Musk admits robotaxis will be geo-fenced and avoid dangerous intersections—after a red light incident exposed major safety flaws in the company’s “Full Self-Driving” tech.
At a Glance
- Tesla’s robotaxis will be geo-fenced in Austin and avoid complex intersections after FSD ran a red light in a media test
- Musk claims the red-light test used a “Supervised” version of FSD, not the “Unsupervised” model intended for robotaxis
- Initial rollout in June will feature just 10 cars, scaling to 1,000 if successful
- Tesla’s system still uses cameras only—no radar or lidar—unlike safety-first competitors like Waymo
- Robotaxis will operate without drivers, but be remotely monitored by Tesla staff
Red Light, Red Flags
It all started with a Business Insider test comparing Tesla’s FSD with Waymo’s fully autonomous vehicles. During the test, Tesla’s system blew through a red light in San Francisco—a fundamental error for any so-called “autonomous” system.
Rather than acknowledge a design flaw, Elon Musk claimed the test “made no sense” because it involved the “FSD Supervised” version. He says the upcoming robotaxi fleet will run a more advanced “Unsupervised” version. But the damage was done: Musk is now embracing geo-fencing and limiting Tesla’s first autonomous rollout to the safest parts of Austin.
A Toned-Down Rollout
Instead of launching a nationwide autonomous fleet, Tesla will debut with just 10 robotaxis in June. The goal is to scale to 1,000 if all goes well. “It’s prudent for us to start with a small number,” Musk conceded, “confirm that things are going well, and then scale it up.”
These Model Y robotaxis will be camera-only, without the radar or lidar used by competitors like Waymo, which already operate in San Francisco and Phoenix without such restrictions. And while Tesla’s vehicles won’t have human safety drivers, they will be monitored remotely by employees watching live feeds.
Watch a report: Tesla Robotaxi Plan Walks Back Full Autonomy.
Vision-Only or Vision-Blind?
Tesla’s insistence on a camera-only approach has long divided experts. Lidar and radar provide vital depth and redundancy—especially critical in edge cases like pedestrian crossings or bad weather. But Musk has long rejected these tools as “crutches,” insisting his vision-only system will be safer and more scalable.
But this latest retreat suggests Tesla’s tech isn’t yet robust enough for unrestricted public use. Avoiding high-risk intersections isn’t just “cautious”—it’s an admission that FSD can’t handle real-world unpredictability, at least not yet.
Politics, Promises, and Pressure
As Tesla’s EV sales slip and investor patience wears thin, Musk is increasingly embroiled in political distractions, including advising Donald Trump and expanding his media footprint. That leaves questions about how focused he remains on Tesla’s core challenges—especially when previous promises about robotaxis have repeatedly failed to materialize.
From a million robotaxis by 2020 to a geo-fenced experiment in 2025, the gap between hype and reality is growing. And with shareholder scrutiny rising, Tesla’s latest pivot may be its most honest move yet: acknowledging that true autonomy still requires human supervision—just from a distance this time.
