Briefing · Regulatory
Why 'Full Self-Driving' and 'Level 4 Autonomous' Are Not the Same Claim
Two phrases that sound similar describe two different regulatory categories, with different legal responsibility attached.
Briefing
"Full self-driving" and "Level 4 autonomous" get used as if they mean roughly the same thing. They don't, and the gap between them is legal, and it has nothing to do with phrasing.
A system marketed under a name implying full self-driving capability can still be, strictly, a Level 2 driver-assistance product. However capable that system is in practice, the law does not care how it feels to use: the driver remains legally responsible and is required to supervise and be ready to intervene at all times, no exceptions carved out because of what the product is called. The name doesn't change the law.
The confusion is easy to understand even if it isn't accurate. Both phrases describe a car handling the driving task, and from inside the vehicle a well-tuned Level 2 system can feel almost indistinguishable, moment to moment, from something operating without supervision. The regulatory categories don't grade on how the drive feels from the seat, though. They grade on who is legally on the hook if something goes wrong, and on that question a Level 2 system's answer never changes no matter how capable the underlying software becomes.
This isn't a hypothetical mismatch, either. The naming gap between a marketed “full self-driving” feature and its actual Level 2 certification has drawn direct regulatory scrutiny in the US, where state regulators have specifically challenged whether marketing a driver-assistance product under a name implying full autonomy is misleading, given that the underlying system remains legally a Level 2 product requiring constant driver supervision. Whatever the outcome of any individual dispute, the underlying argument on both sides turns on exactly the distinction laid out here: what the name implies against what the certification actually requires.
Level 4 is a different category entirely. A Level 4 system is designed and engineered to operate without human supervision, but only within a defined operating domain, a specific geography, road type, weather condition set, or combination of those. Deploying a Level 4 vehicle commercially generally requires the company to obtain a driverless deployment permit from the relevant regulator, a separate and typically much harder authorization than the testing permits that cover earlier development work.
None of this means the naming choice is accidental or careless. A name that gestures toward full autonomy signals a company's long-term product roadmap and technical ambition to customers, investors and competitors alike, even while the certified capability on the road today remains narrower. The trouble isn't that the ambition is stated; it's that a road name and a regulatory category answer different questions, and only one of them determines who's legally responsible if the system fails to react in time.
None of this is a judgment about which system is better engineered or more capable in absolute terms. It's a description of what each regulatory category legally requires of the company and the vehicle, full stop. A Level 2 system with excellent real-world performance is still, legally, a Level 2 system, and a Level 4 permit does not get issued because a product name suggests the car can drive itself. The underlying taxonomy is covered in more depth in the SAE levels explainer, and which companies currently hold which deployment permits, in which cities, is the kind of thing the deployment tracker exists to keep current rather than to explain from first principles.
More Regulatory briefings
All briefings are reference and analysis pieces, distinct from the 2013–2018 news archive.