Self-Certified Does Not Mean Self-Approved
Tesla's Cybercab inquiry exposes the difference between a manufacturer's legal declaration, a regulator's audit and an exemption for a vehicle built without human controls.
In short
What happened. The U.S. auto-safety regulator is auditing Tesla’s certification that its steering-wheel-free Cybercab meets every federal safety standard that applies to it.
What it means. American vehicle certification begins with the manufacturer, not with advance government approval. The regulator can test that declaration afterward, and the Cybercab is now that system’s unusually visible test.
Risks and impact. The inquiry does not establish that the vehicle is unsafe or noncompliant. It asks how rules written around a human driver apply when the car has no wheel, pedals or mirrors.
What can be done. A rider can separate four questions: whether a vehicle is certified, whether it has an exemption, where and when it may operate, and what public evidence supports its safety claims.
What to watch. Tesla’s answers, due September 30, should clarify whether temporary human controls helped form the basis of its certification.
Shown as a summary because of your reading settings.
What happened
Tesla began commercial deployment of a small number of two-seat Cybercabs in Austin, Texas, on September 3. The vehicles lack a conventional steering wheel, brake pedal, accelerator pedal and mirrors, according to the National Highway Traffic Safety Administration and Reuters.
On September 4, NHTSA opened an Audit Query into Tesla’s declaration that the Cybercab complies with all applicable Federal Motor Vehicle Safety Standards, or FMVSS. On September 15, Reuters reported that the agency had ordered Tesla to answer a series of questions by September 30. They include whether temporary human-driving controls or other equipment contributed to the compliance case, whether a person can drive the vehicle, whether occupants can move it through touchscreen controls, and what limits apply to speed, geography and time of day.
This is an inquiry, not a finding. NHTSA has not said that the Cybercab failed a standard, and Tesla did not respond to Reuters’ request for comment. The agency says existing rules remain in force while it works on several changes for automated vehicles. Its March proposal to remove a gear-display requirement for vehicles without manual controls was still a proposal, not a blanket waiver for every driver-oriented rule.
What the evidence supports
The public record supports three firm conclusions. First, the United States uses manufacturer self-certification. NHTSA writes performance standards; a maker declares that its vehicle meets the applicable ones; the agency then retains authority to check compliance and demand a remedy. “Self-certified” therefore describes who makes the first legal declaration. It does not mean the declaration is beyond review.
Second, the design creates genuine classification questions. NHTSA says its audit will examine the technical data and processes Tesla used, including any conclusion that a requirement did not apply. Reuters’ account of the agency’s questions adds the specific issue of temporary manual controls. Neither document supplies Tesla’s answers, so it is too early to judge the reasoning.
Third, there is an established alternative when compliance with a particular rule would prevent a novel vehicle from operating. In July, NHTSA granted Zoox a two-year exemption from portions of eight standards, allowing no more than 2,500 exempt vehicles into commercial deployment in any 12-month period. That decision required an equivalent overall level of safety and continuing operational conditions. It shows what an exemption looks like; it does not prove that every unconventional vehicle needs the same one.
How the story is being framed
One framing calls this a loophole story: a company has put an unfamiliar machine on public streets by certifying its own work. That captures the discomfort of seeing the declaration precede the audit. It misses that self-certification is not a Tesla exception. It is the ordinary U.S. system, backed by recalls, civil penalties, defect investigations and post-market testing.
A second framing treats the inquiry as evidence that old rules are blocking useful technology. There is truth in the mismatch. A gear indicator meant to prevent a driver from selecting the wrong position has little obvious purpose in a vehicle nobody can drive. NHTSA made exactly that argument in its March proposal. But “written for humans” does not make every requirement pointless: lighting, braking, crash protection and emergency exit still protect passengers and everyone outside the car.
A third framing reads the audit as proof that Cybercab is unsafe. The released documents do not support that conclusion. Compliance and automated-driving performance also are different questions. A vehicle can satisfy equipment rules yet make poor driving decisions; it can also need a narrow regulatory exemption while demonstrating an equivalent safety level.
The useful frame is less dramatic: this is a boundary test between rules, evidence and product design. Tesla’s response may show a defensible mapping of each standard, or expose gaps that require correction. Until the record contains that response and NHTSA’s assessment, certainty in either direction is borrowed confidence.
The background
Most consumer products do not arrive with a regulator standing at the factory door to approve every unit. Cars in the United States follow a particularly important version of that model. Federal standards specify measurable requirements, manufacturers attach certification labels, and NHTSA polices the market through information requests, testing, investigations, recalls and enforcement. That architecture can scale across many models, but it depends on accurate declarations and credible oversight.
Automated vehicles strain it because many FMVSS provisions were drafted with a driver in mind. Since 2018, NHTSA has been asking how to remove regulatory barriers without discarding the safety purpose beneath them. It revised occupant-protection rules in 2022 and this year proposed an exception to the transmission-shift display for an automated vehicle with no manual controls. The proposal carefully preserved the existing requirement for an automated vehicle that still has controls.
Exemptions provide a bridge while permanent rules catch up. The July Zoox order covered portions of standards concerning defrosting, wipers, lamps, rear visibility, brakes, interior impact, glazing and occupant protection. It was bounded by time, volume, labeling and operational oversight. The order also states the broader rule: an automated vehicle is not prohibited from public roads if its manufacturer certifies compliance with all applicable standards and the system does not present an unreasonable safety risk.
That leaves two separate layers of governance. Federal law largely governs vehicle construction and equipment. State and local authorities govern much of road operation, licensing and service geography. A city allowing rides does not certify the vehicle’s federal compliance; a federal certification does not tell a passenger the service’s current operating domain.
Who it touches
NHTSA Administrator Jonathan Morrison described the agency’s task as balancing automated-vehicle innovation with safety oversight. That phrase sounds abstract until a vehicle arrives at the curb with no obvious place for a human to take over.
For a passenger, the practical questions are immediate: Can the vehicle operate only in mapped parts of Austin? What happens outside its design limits? How does someone stop or exit it? Who responds remotely? For a cyclist or pedestrian, there is no consent screen at all; the system’s behavior is simply part of the street.
The regulator’s questions about speed, geography and time are therefore not side details. They define the operating design domain—the conditions in which the automated system says it can do the whole driving task. A safety claim without those boundaries is like a weather forecast without a place or date: technically shaped like information, but hard to use.
The deeper story
Certification is often mistaken for a medal. In practice it is closer to a signed statement with consequences. The distinction matters far beyond robotaxis. A label can tell us who accepts legal responsibility and which test applies. It cannot compress every uncertainty about a new technology into one word.
The Cybercab case exposes a recurring problem: rules preserve lessons from earlier failures, while invention rearranges the objects those rules describe. A mirror requirement assumes a driver who looks backward. Remove the driver, and the literal object may become obsolete. The protected function—awareness of what is behind and beside the vehicle—does not. Good regulation must know which part is furniture and which part is memory.
That is why neither “the rule says so” nor “the technology is new” is a complete argument. The first can freeze an old interface after its safety purpose has migrated elsewhere. The second can erase hard-won protections by treating novelty as evidence of equivalence. The more honest question is functional: what harm was this rule designed to prevent, and what verifiable mechanism prevents that harm now?
For a reader evaluating any automated service, a four-part check is more useful than a single badge. Ask what was self-certified; what, if anything, was exempted; what operating limits are published; and what evidence or reporting can be inspected independently. Those answers will not predict every failure. They do reveal whether “safe” is being used as a conclusion, a process or merely a promise.
Something to sit with
When a familiar safety device disappears, what evidence would convince you that its function still exists?
Which matters more to your trust: who makes the first certification, or how quickly a weak certification can be found and corrected?
Sources
- National Highway Traffic Safety Administration — https://www.nhtsa.gov/press-releases/investigation-tesla-cybercab-s...
- Reuters — https://www.reuters.com/business/autos-transportation/us-agency-ord...
- Federal Register — FMVSS No. 102 proposal — https://www.federalregister.gov/documents/2026/03/16/2026-05024/fed...
- Federal Register — Zoox temporary exemption — https://www.federalregister.gov/documents/2026/07/31/2026-15485/zoo...
We report facts from the sources above in our own words and link to the originals. Interpretation is ours, not theirs.
Who initially certifies that a new vehicle meets applicable U.S. federal motor vehicle safety standards?
Manufacturers make the certification; NHTSA sets the standards and can audit, investigate, order remedies or grant limited exemptions.
♻︎ Free to republish
Copy this HTML into your CMS. Credit line and licence are included. Republish our work — free
Every headline has a deeper story. This is ours.
What we are doing here