How Donald Trump could help Elon Musk with his robotaxi plans
Tesla's Cybercab has no steering wheel or pedals—that may be illegal under current law.
I’m helping to organize the Ride AI conference, which will be in Los Angeles on April 2, 2025. If you’re a leader in the autonomous vehicle industry I hope to see you there. Click here to get more details and purchase a ticket.
Yesterday Bloomberg reported that the Trump transition team plans “to make a federal framework for fully self-driving vehicles one of the Transportation Department’s priorities.” Such a framework would be beneficial to Elon Musk, who aims to introduce a driverless robotaxi called the Cybercab as early as 2026.
While Bloomberg had an important scoop, I think their writeup may be confusing to people who aren’t experts on federal automotive safety law. So here’s a brief explainer on the current state of the law and what the Trump administration—or the new Republican Congress—might change over the next two years.
At the federal level, automotive safety is governed by the National Highway Traffic Safety Administration, which has written regulations called the Federal Motor Vehicle Safety Standards (FMVSS). The FMVSS governs almost every safety-related aspect of vehicle design: brakes, seat belts, headlights, windshields, and so forth.
The FMVSS is performance-based and relies on self-certification. For example, if I’m reading FMVSS standard 105 correctly, a passenger car must be able to brake in less than 150 feet from a starting speed of 50 miles per hour. So before an automaker can introduce a new car model, it must take the vehicle out on a test track, verify that its brakes meet this standard, and certify to NHTSA that it has done so. And it must do similar testing on many other aspects of the car’s performance.
The FMVSS, which has been around since the 1960s, doesn’t have any specific rules for self-driving cars. But it assumes that cars will have human drivers—and hence steering wheels, accelerator and brake pedals, side and rear-view mirrors, and so forth.
In some cases the regulations don’t explicitly require these things—they simply assume they will exist and mention them in related performance standards. Still, it’s probably not possible for a vehicle with no steering wheel or pedals to comply with the FMVSS as currently written. And that’s a problem for companies that are planning to build and deploy such vehicles.
A self-driving company can sidestep this issue entirely by building atop a conventional vehicle. Waymo, for example, has been purchasing I-PACE cars from Jaguar and modifying them to be fully self-driving. Because the base vehicle already complies with the FMVSS, Waymo’s robotaxis are automatically compliant. The downside for Waymo is that the driver’s seat has to be kept empty during a ride, reducing the capacity of its robotaxis.
By the same token, Tesla hasn’t had trouble certifying the Model 3 or Model Y as FMVSS-compliant. Because these were primarily designed as human-driven vehicles, they have steering wheels, pedals, mirrors, and other required equipment. That won’t change if and when Tesla pushes out a software update that makes them capable of full self-driving.
So the question here isn’t whether the federal government legalizes driverless vehicles—those are already permitted at the federal level. The question is whether to remove the de facto requirement that all vehicles have steering wheels and pedals—requirements that make fully driverless vehicles less useful by effectively requiring them to have a vestigial driver’s seat.
Four strategies for deregulation
If the Trump Administration wants to allow driverless vehicles with no steering wheels or pedals, they’ll have four basic options.
First NHTSA could rewrite the FMVSS. These rules were not written by Congress. Rather, Congress gave federal regulators broad authority and instructed them to flesh out specific safety standards. So the new NHTSA administrator could initiate a rulemaking process to remove rules requiring vehicles to have steering wheels and pedals.
But that process is likely to be slow. Rulemaking requires a notice-and-comment process that takes at least a year and probably two. And there’s a risk that automotive safety advocates would sue, arguing that the new rules aren’t consistent with the guidelines set by Congress. That could easily delay the rules by a couple more years, preventing them from taking effect before the end of Trump’s term.
It doesn’t help that a landmark Supreme Court ruling this year overturned a principle called Chevron deference that said that courts should defer to regulatory agencies in their interpretation of ambiguous statutes.
Elon Musk wants to start selling the Cybercab in 2026, so a process that could get bogged down in litigation beyond 2028 isn’t going to be attractive to him.
Second, NHTSA could grant waivers from NHTSA rules. NHTSA has the power to grant exemptions from the FMVSS on a case-by-case basis. In 2020, the agency did that for Nuro, a startup that makes delivery robots.
Back in 2022, GM’s self-driving subsidiary, Cruise, asked for an exemption for the Origin, a custom-designed robotaxi with no pedals or steering wheel. NHTSA never acted on the request. Cruise ultimately canceled the Origin after one of its conventional robotaxis (based on the Chevy Bolt) struck and dragged a pedestrian in San Francisco. So NHTSA never formally made a decision on the exemption request.
Even if the second Trump Administration grants these exemptions more freely than the Biden Administration has, they have a big downside: each exemption is limited to 2,500 vehicles per year. That cap is unlikely to be a problem in the first few years of a robotaxi service—Waymo still has fewer than 1,000 commercial vehicles four years after launching its robotaxi service. But it probably won’t be good enough for Tesla, which is presumably hoping to sell more than 2,500 Cybercabs in the first year after they are introduced.
Third, NHTSA could simply look the other way as companies certify their own compliance with the FMVSS.
The Amazon-owned startup Zoox has been in a standoff with NHTSA for over a year over its own custom driverless vehicle, which has no steering wheel or pedals. In 2022, Zoox asserted that its vehicles are compliant with the FMVSS. Last year NHTSA opened an investigation into Zoox’s decision to certify its vehicles, perhaps signaling that the agency does not think the vehicles are compliant. That investigation remains open as Zoox prepares to start using the vehicles in San Francisco and Las Vegas.
It’s possible the Trump administration could simply drop the Zoox investigation and signal to other companies that it will not second-guess their own certification decisions. But companies that take this route could get in legal trouble if a more regulation-minded administration takes over in 2029 and decides these companies have been flouting the law.
Elon Musk has a high tolerance for legal risk, so I could imagine him going this route. But he would presumably prefer to be on more solid legal ground.
Congress could change the rules
That brings us to the fourth option, which is that Congress could pass new legislation. According to Bloomberg, “a bipartisan legislative measure being discussed in early stages would create federal rules around AVs.” Bloomberg adds that “policy details have yet to be determined.”
Legislation like this has been floating around Capitol Hill for the better part of a decade. One of the most significant efforts was in 2018, when Congress considered legislation—dubbed SELF DRIVE in the House and AV START in the Senate—that instructed the Secretary of Transportation to revise the FMVSS to allow operation of autonomous vehicles.
Recognizing that the process would take several years, these bills temporarily raised the annual exemption cap from its current level of 2,500. The House bill would have allowed exemptions for up to 275,000 vehicles over four years, whereas the Senate bill would have allowed 215,000 vehicles over four years.
Ultimately neither bill became law, and as far as I can tell there have been no meaningful changes to the rules over the last six years.
I remember feeling ambivalent about this legislation in 2018, and I expect to feel similar ambivalence about new legislation. I’m a fan of self-driving technology and do not want regulators to interfere with its deployment. However, current law allows companies to deploy an unlimited number of driverless vehicles with steering wheels, as Waymo has been doing since 2020. It’s not clear that the steering wheel requirement is a significant impediment to scaling up this technology.
At the same time, there are real safety benefits to self-driving cars having steering wheels. There were a number of incidents last year when Waymo or Cruise vehicles got confused at the scenes of car crashes, fires, or other emergency situations. One way to deal with these situations is for a police officer or firefighter to hop into a vehicle, take manual control, and drive the car where it needs to go.
I expect this to become less necessary over time. But it doesn’t seem like such a bad idea to keep steering wheels in self-driving cars for at least the first few years.
If Congress removes the steering wheel requirement—or instructs NHTSA to do so—that’s likely to be paired with a new set of safety standards. I can imagine a Republican Congress preempting strict autonomous vehicle regulations in blue states like California and New York.
The problem is that I’m not sure we know what federal standards should look like. The technology is still evolving fast, and any laws that get passed now might quickly become obsolete. A few more years of hands-on experience with the technology might equip policymakers to write clearer and more effective standards.
But the case for urgency is clear for Musk, who has staked a lot on the success of the Cybercab. The lack of a steering wheel may make the Cybercab illegal under current federal regulations.
"At the same time, there are real safety benefits to self-driving cars having steering wheels. There were a number of incidents last year when Waymo or Cruise vehicles got confused at the scenes of car crashes, fires, or other emergency situations. One way to deal with these situations is for a police officer or firefighter to hop into a vehicle, take manual control, and drive the car where it needs to go... [this will] become less necessary over time. But it doesn’t seem like such a bad idea to keep steering wheels in self-driving cars for at least the first few years."
I understand the appeal of incrementalism, and the value for safety. I do!
But there is a hidden cost with keeping the steering-wheel requirement, which is it locks in "automated driving as a feature for personal vehicles" as the default. The critical path that sets up is one where most people continue to own their own car. If we want other kinds of automated driving—not merely robotaxis, but automated delivery vehicles, automated transit—we need to permit experimentation with form factor... not as a 'nice to have' in the future, but as part of the first wave.
Not for nothing is Zoox (and Tesla with its Cybercab) coming hard out of the gate with a custom vehicle; Cruise wanted to do the same. They think, and I believe correctly, that robotaxi in a conventional vehicle will fail to achieve the value necessary to overcome traditional automobility.
Musk has a bigger problem than whether the lack of a steering wheel makes the car illegal. He'd have to prove his car is safe without a driver.
Cutting corners with self-driving cars does not work, which Cruise found out the hard way. Tesla just doesn't have what it takes in terms of its tech, unlike Waymo.