The UK's Automated Vehicles Act and What It Covers

Royal Assent landed on the UK's Automated Vehicles Act in 2024, and with it came something most driverless programs elsewhere have had to build around rather than rely on: a statutory answer to the liability question. Once a vehicle is authorised as self-driving under the Act, legal responsibility for its driving shifts away from the human occupant. That single mechanism, allocating liability by statute rather than leaving it to case law or insurance contract terms, is the Act's most consequential provision. That's rare in this space.

In practice, that shift changes who a claimant sues after a collision. Under the older default, a human occupant's own liability and insurance were the first port of call regardless of how much control they actually had over the vehicle in the moment. Once a system is authorised as self-driving under the Act, that first port of call moves to whichever party the statute designates instead, without requiring a court to work out from first principles who was really in control when nobody was steering.

The Act also sets a specific benchmark for what “authorised” means in the first place: a vehicle has to demonstrate a standard of safety broadly equivalent to, or better than, a careful and competent human driver before a regulator will authorise it as self-driving at all. That is a comparative standard rather than an absolute one, it doesn't require perfection, it requires performing at least as well as the standard the law already expects of any driver on the road. Setting the bar there, rather than at some fixed numerical target, avoids one problem while creating another: it keeps the standard tied to something courts and regulators already understand, but it also means the careful-and-competent-driver baseline itself has to be argued and evidenced case by case rather than read off a spec sheet.

The Act assigns responsibility to a specific role it calls the authorised self-driving entity, the party responsible for the vehicle's self-driving behaviour once it's authorised. That's a deliberate design choice: rather than leaving the question open to whichever party happens to be closest to a given deployment, whether that's a vehicle manufacturer, a software developer or a fleet operator, the statute forces one party to be named as answerable for the driving itself before the vehicle can operate at all.

Most jurisdictions have arrived at driverless authorization piecemeal, through permits, pilot exemptions and agency rulemaking that accumulate over years without ever being codified into a single law a court can point to. The UK chose the opposite path: one Act, one national authorisation process, applied the same way regardless of which city a service operates in. A company authorised under the Act does not need a separate sign-off from each local authority the way it would under a permit-by-permit system.

This puts the UK in the same category as Germany's national Level 4 statute, and the contrast both countries draw is with the United States, where no equivalent federal framework exists and each state runs its own permit regime. The structural question is the same one raised in the Germany comparison: does legislating nationally actually move deployment faster, or does it just centralize the paperwork? The Act is still young enough that the answer isn't settled.

What it has settled, at least on paper, is who answers for a crash when nobody was driving.