Tesla Cybercab is barely real and it’s already under investigation

The Verge’s story (see https://www.theverge.com/transportation/990130/tesla-cybercab-nhtsa-investigation-fmvss-certify) captures an almost inevitable next strand in Tesla’s saga—a product that barely exists, but already attracts the kind of regulatory attention that most startups can only imagine. The NHTSA’s audit query (AQ) is not a standard safety recall; it’s an inquiry into how Tesla intends to validate the Cybercab’s compliance with Federal Motor Vehicle Safety Standards (FMVSS) when the design intentionally flouts long-standing assumptions about what a car driver ought to be. The question isn’t whether the car can be safe, but whether the process used to get there can be trusted. This is very Tesla: a company that often treats litigation and regulation as speed bumps, not boundaries. But there’s a deep tension here. Tesla's previous crash issues, like Autopilot, already forced NHTSA into closer investigation—yet those inquiries were allegations about software overpromises. The Cybercab is engineered to delete the human fallback entirely, and that raises a more fundamental existential question: should we even be certifying a vehicle whose entire safety case is built around a still-rolling-out “full self-driving” system? The government is not saying the car is unsafe; it is asking whether the process to show that fits a legal framework written for physical human controls like steering wheels, side-view mirrors, and pedals. The broader theme is the collision between a software-first view of transportation and a hardware-centric legal system. Musk’s entire robotaxi strategy is to skip all iterative validation steps and use a “production as proof” approach. yet NHTSA' step does underscore that a federal legal body can’t just wave a magic wand and vaporize a legal framework, even when a billionaire talks about “safety eventually.” For years, industry leaders have argued that autonomous vehicles should be evaluated against a baseline of human driving’s error rates, and that rigid FMVSS for human infiltration may actually slow safety progress. The Cybercab might be the perfect test case for that argument—but that test case is happening in front of a regulator that has to adhere to the letter of the law, not the speculative promise. For now, the Cybercab is a prototype, a gilded concept with a regulatory headache before it even has a commercial launch date. The story rare given the mystique of a headline. But it should later be a warning: the hardest part of making robot cars isn’t the computer vision, it’s the process of remaking the world—to in a world where the default assumption is a human behind the wheel. As The Verge’s report hints, there is no firm “to examine the process and [the certification]”—for the Cybercab, the future is not high-ceilinged, it’s administrative. And if that’s what’s holding back autonomy, we’re all about to learn how to love the audit query.