
When people hear self driving cars, they often picture very different things. For some, it’s a Tesla using Autopilot on the motorway. For others, it’s a Waymo robotaxi picking up passengers without a steering wheel in sight. This confusion is exactly why SAE International created its levels of driving automation, a six-step scale that defines how much control the car has versus how much the driver must retain.
These levels, set out in the SAE J3016 standard, are now used worldwide by automakers, regulators, and insurers. They matter because they don’t just describe technology, they set expectations for safety, responsibility, and liability. Understanding them is key if you want to know how close we are to a truly driverless future.
What Are the SAE Levels of Automation?
SAE levels describe the degree to which a car can perform the driving task. At one end of the spectrum, Level 0 means the driver does everything, while Level 5 means the car drives itself under all conditions. Each step up represents more tasks handled by the vehicle’s system and less responsibility on the driver.
Governments, safety agencies, and insurers have adopted this framework because it makes discussions about autonomy consistent. For example, when the UK introduced the Automated Vehicles Act 2024, which sets safety standards for autonomous cars, the SAE levels were the benchmark behind the legal definitions (gov.uk).

Level 0: No Automation
Level 0 means no automation in the driving task. The driver controls everything, though the car may provide warnings like lane departure alerts or blind spot detection. These alerts never act on the vehicle's controls.
This is the level most UK drivers were used to until the 2010s, before advanced driver assistance systems (ADAS) became widespread.
Level 1: Driver Assistance
Level 1 is the first step toward automation. The car can control either steering or speed, but never both at the same time. Adaptive cruise control is a common example, adjusting your speed to maintain distance from the car ahead while you continue to steer.
Level 1 systems sit firmly under the driver's supervision. ADAS features like adaptive cruise are widely credited with reducing rear-end collision rates, but they don't shift responsibility from the driver.
Level 2: Partial Automation
At Level 2, the vehicle can handle both steering and speed simultaneously under certain conditions. The driver must stay fully engaged and ready to intervene at any moment.
This is the most advanced automation available in consumer cars today. Tesla's Autopilot, Ford's BlueCruise, and GM's Super Cruise all fall into this category. They can keep a car centred in its lane, change lanes on command, and adjust speed. If the system encounters poor weather, unexpected roadworks, or an unclear situation, the driver must instantly take over.
This is where confusion often arises. Drivers sometimes assume Level 2 systems are "self-driving", leading to overconfidence. Regulators stress that at this stage the driver is still legally responsible for the car.
Level 3: Conditional Automation
Level 3 is the first stage where a car can truly drive itself, but only in specific, pre-defined conditions. The system monitors the environment, makes driving decisions, and doesn't require the driver to constantly watch the road. However, it can request the driver retake control with little warning if conditions change.
Mercedes-Benz has pioneered this with its Drive Pilot system, approved for use in Germany and in certain US states (California and Nevada). In slow-moving motorway traffic, drivers can legally take their eyes off the road for tasks like checking emails, until the system alerts them to resume control.
The challenge here is liability. If a crash occurs while the vehicle is driving itself, does the fault lie with the driver, the automaker, or the software provider? This grey area is part of why the UK's Automated Vehicles Act 2024 was brought in, and why insurers are already adapting policies for higher-level autonomy. Our explainer on how Zego is insuring the autonomous revolution walks through how UK insurers are preparing for the Level 3+ liability shift.
Level 4: High Automation
Level 4 vehicles can drive themselves without human input, but only in specific areas or under certain conditions. This is known as geofenced autonomy.
Waymo, for instance, operates Level 4 robotaxis in cities such as Phoenix and San Francisco. Passengers hail a ride without touching the controls, and no safety driver is required. These cars are restricted to mapped urban zones and favourable weather conditions.
Level 4 is where automation starts to reshape transport at scale, particularly for commercial fleets and shared mobility. Industry analysts have forecast the global autonomous vehicle market could reach several hundred billion dollars by 2035, driven largely by fleet services and robotaxis.
For a wider look at whether the technology actually lives up to the safety claims, our piece on are self-driving cars really safer than humans breaks down the evidence.
Level 5: Full Automation
Level 5 is the ultimate goal. Full autonomy in all conditions, everywhere. At this stage, cars wouldn't need a steering wheel, pedals, or human involvement at all.
Despite years of hype, no Level 5 vehicles exist today. Even the most advanced systems are still constrained by weather, road complexity, or geography. Experts believe it could be decades before true Level 5 autonomy is achieved, if it ever is. For a deeper look at the technology in scope, our guide on what self-driving cars actually are walks through the core components.

Why the SAE levels matter
The SAE levels are more than a technical curiosity. They provide a shared framework for how society approaches autonomy.
- For drivers, they explain what the technology can and cannot do.
- For manufacturers, they help market features responsibly without overstating autonomy.
- For regulators, they form the backbone of laws like the UK's Automated Vehicles Act 2024.
- For insurers, they clarify where liability sits as cars take on more of the driving task.
Trust and accountability matter as much as the technology itself. The SAE levels are the roadmap that makes this transparent.
Where we stand in 2026
The vast majority of cars on UK roads are Level 1 or Level 2. A handful of Level 3 systems are being trialled in specific markets. Level 4 robotaxis are live in a few US cities. Level 5 remains theoretical.
Progress is gradual. Consumer demand for advanced automation features is strong, but widespread adoption depends on proving safety, lowering costs, and resolving liability questions. The UK's AV Act 2024 is the first major legislative step to clarifying how that liability will be handled as higher-level cars reach British roads.
What this means for insurance
The SAE levels are reshaping UK car insurance from the underwriting up. At Level 2 and below, the driver is still legally responsible, which is how most motor insurance is priced. At Level 3 and above, liability starts to shift toward the vehicle manufacturer and software provider, which is what the UK AV Act 2024 codifies.
In the meantime, telematics-based insurance uses the same principle that will eventually price autonomous cars: real driving data, not demographic assumptions. Telematics car insurance that rewards safer driving from Zego measures how you actually drive through the Zego Sense app, and safer drivers typically pay less at renewal regardless of age or postcode. The same behavioural-data principle will underpin AV insurance when Level 3+ vehicles are priced for UK roads.
Frequently asked questions
What's the difference between Level 4 and Level 5 autonomy?
Level 4 cars can operate without human input, but only in specific geofenced areas or conditions. Robotaxis from Waymo are Level 4 because they operate in mapped city zones but can't drive everywhere. Level 5 represents full autonomy in all conditions, with no steering wheel, pedals, or human backup. No vehicles have reached Level 5 today.
Are self-driving cars already on UK roads?
At a limited level, yes. Most consumer vehicles available today are Level 2, with partial automation like lane keeping and adaptive cruise control. Level 3 systems are just beginning to appear in select models and markets. Level 4 robotaxis are being tested and operated in some US cities, with UK trials under the AV Act 2024 framework, but widespread Level 4 adoption in the UK is still years away.
Who is responsible if a self-driving car crashes?
Responsibility depends on the level of automation. At Level 2, the driver is always accountable because the car is only assisting. At Level 3 and above, liability becomes less clear. Is it the driver, the manufacturer, or the software provider? This legal question is exactly what the UK's Automated Vehicles Act 2024 was introduced to clarify, and why insurers are developing new AV-specific cover.
When will we see fully autonomous cars in the UK?
Industry forecasts suggest gradual progress rather than a sudden leap. Some analysts expect Level 3+ features in a meaningful share of new cars by the early 2030s, with Level 4 fleets scaling first in cities. Level 5, where cars can drive anywhere without human input, may still be decades away, and some commentators argue it may never be practical for every road condition.
Why do the SAE levels matter to UK drivers?
The SAE levels help set clear expectations. A car with Level 2 features is not a "self-driving car", even if it looks advanced. Knowing the difference helps drivers stay safe, prevents misuse, and informs buying decisions. It also shapes how UK insurers and regulators treat the technology, so understanding the levels means understanding your own responsibilities behind the wheel.