Driverless Taxis Have Arrived. But Who Pays if They Crash?

Driverless Taxis in London: Who's Responsible When Something Goes Wrong?

Londoners can now find themselves being driven through the capital by a car that's doing much of the work itself.

On 3 September 2026, Uber and British autonomous vehicle company Wayve launched supervised autonomous rides in London, using electric Ford Mustang Mach-E vehicles equipped with automated driving technology.

Passengers booking an UberX, Uber Electric or Uber Comfort journey may be matched with one of these vehicles at no additional cost. They can accept the autonomous ride or switch to a conventional Uber.

But there's an important detail: these aren't fully driverless taxis.

Every journey has a trained, Transport for London-licensed private hire driver in the driving seat, supervising the system and ready to take control if required.

It's a significant step towards autonomous transport, but it's not quite the world of completely driverless taxis just yet.

For motorists, passengers and insurers, the development raises some interesting questions.

Can autonomous vehicles really make our roads safer? Who's responsible when software makes a mistake? And could driverless technology eventually make motor insurance cheaper?

At Performance Direct, we arrange insurance for a wide range of vehicles, including performance, modified and specialist cars. As vehicle technology develops, understanding how it changes the risks involved will become increasingly important.

Let's look at what London's new autonomous taxis mean for the future of motoring and insurance.

Why is London such an important testing ground?

If you're going to test whether a car can handle unpredictable driving situations, London certainly offers plenty of opportunities.

Busy junctions, cyclists, pedestrians, roadworks, narrow streets and drivers making last-minute decisions are all part of the daily experience.

An autonomous vehicle needs to recognise these hazards, predict what might happen next and respond safely.

That's quite a challenge when the vehicle ahead suddenly changes lanes or a pedestrian steps into the road.

Wayve has been developing and testing its technology in London since 2018.

Its system uses artificial intelligence that learns from driving experience, rather than relying entirely on detailed maps or individually programmed rules for every possible situation. The company says this approach is intended to help its technology adapt to different driving environments.

What could autonomous taxis offer passengers?

The potential benefits extend beyond the novelty of travelling in a car that's driving itself.

Autonomous taxi services could eventually provide more independence for people who cannot drive, including some older and disabled passengers.

They might also offer an alternative to private car ownership, particularly in cities where parking is expensive and public transport doesn't always meet everyone's needs.

However, the benefits will depend on how reliably the technology performs, whether the services are accessible and affordable, and how much confidence passengers have in their safety.

And that's where the questions become more complicated.

It's one thing for a car to complete a journey without incident.

It's quite another to establish what happens when something goes wrong.

Are autonomous cars actually safer than human drivers?

Possibly, but there isn't a universal answer yet.

Different autonomous systems operate in different environments, with different technological capabilities and operating restrictions.

Evidence gathered on American highways, for example, cannot automatically tell us how a particular system will perform on narrow British streets during heavy rainfall.

There are good reasons to investigate the potential safety benefits.

Government analysis found that human error was recorded as a contributing factor in 88% of road collisions in Great Britain in 2020.

Automated driving could potentially reduce some risks associated with tiredness, distraction and impaired driving.

A computer doesn't become impatient, decide to answer a phone call or attempt to drive home after a few drinks.

However, removing some human behaviours doesn't eliminate every risk.

Autonomous vehicles introduce different challenges.

Sensors can be affected by weather and visibility. Software might misinterpret an unusual situation. Digital systems can experience faults or become targets for cyberattacks.

They must also interact with human drivers, whose next move isn't always predictable.

What safety standards will autonomous vehicles need to meet?

The Automated Vehicles Act 2024 provides a framework for the authorisation and regulation of self-driving vehicles in Great Britain.

The legislation requires safety principles designed to ensure that authorised automated vehicles achieve a level of safety equivalent to, or higher than, that of a careful and competent human driver.

The principles must also aim to improve overall road safety in Great Britain.

That establishes a demanding objective.

However, demonstrating that a vehicle meets the required standard will depend on reliable evidence from testing and real-world operation.

Regulators will need to understand how systems perform in different weather conditions, on unfamiliar roads and in unusual traffic situations.

The same evidence will also be important for insurers when assessing the risks.

Who's liable if an autonomous car causes an accident?

This is where things become particularly interesting for motor insurance.

With a conventional vehicle, an insurer will usually investigate what the drivers involved were doing.

Was someone speeding? Were they distracted? Did they fail to give way?

But what happens when the vehicle's software was performing the driving task?

An accident could involve a sensor malfunction, a software error, inadequate maintenance or the actions of another road user.

Establishing responsibility may require a very different investigation.

How does UK law deal with autonomous vehicle accidents?

The Automated and Electric Vehicles Act 2018 began addressing this issue.

Under the Act, where a listed automated vehicle is insured, is driving itself and causes an accident on a road or other public place in Great Britain, its insurer can be liable in the first instance for the resulting damage, subject to the statutory requirements.

The legislation also contains specific owner-liability provisions for certain vehicles that are legally exempt from compulsory insurance. It does not automatically make every owner of an uninsured automated vehicle liable under those provisions.

The intention is to provide an injured person with a route to compensation without requiring them to establish immediately whether the fault originated with a manufacturer, software supplier, sensor or another organisation.

The insurer may subsequently have rights to recover costs from another party responsible for the accident.

In other words, the person making the claim shouldn't have to resolve a complicated technical dispute before beginning the compensation process.

There are qualifications.

Compensation may be reduced where an injured person contributed to the accident.

The legislation also contains provisions concerning prohibited software alterations and failures to install safety-critical updates, which can affect the rights and liabilities of those involved.

The circumstances and applicable legislation will determine how each claim is assessed.

What happens when there's nobody in the driving seat?

London's current autonomous Uber service still has a supervising driver.

However, the next stage of development could involve vehicles completing entire journeys without anyone responsible for driving sitting behind the wheel.

That's where the Automated Vehicles Act 2024 becomes particularly important.

The legislation establishes a framework that distinguishes between responsibility for the automated driving system and responsibility for operating the vehicle.

It introduces several important roles.

The authorised self-driving entity

Every authorised automated vehicle will need an authorised self-driving entity, or ASDE.

This will normally be the manufacturer or developer responsible for the automated driving technology.

The organisation will be responsible for ensuring that the vehicle continues to meet the required self-driving standards and complies with the relevant regulatory obligations.

The no-user-in-charge operator

A vehicle capable of completing an entire journey without a responsible person in the driving seat will also need a licensed no-user-in-charge operator.

This organisation will oversee responsibilities outside the driving task.

These include maintaining the vehicle's roadworthiness, responding to breakdowns and managing certain operational responsibilities.

The business responsible for developing the driving technology and the organisation operating the vehicle may be different companies.

What about the person travelling inside?

For authorised vehicles that can drive themselves for part of a journey but may need someone to resume control, the legislation introduces the role of a user-in-charge.

When an authorised self-driving feature is engaged and operating within its approved conditions, that person will generally not be responsible for offences arising solely from the way the vehicle drives.

However, they retain responsibilities outside the automated driving task.

They may also need to respond appropriately when the system issues a valid request to resume control.

For example, they cannot simply ignore a request to take over when the vehicle reaches the limit of its authorised operating conditions.

This is very different from conventional driver-assistance systems, where the driver remains responsible for supervising the vehicle.

When will the new framework be fully implemented?

The Government is still developing the secondary legislation and detailed safety requirements required to implement the Automated Vehicles Act 2024.

Full implementation is currently intended for the second half of 2027.

Until then, supervised services such as the Uber and Wayve launch represent an intermediate stage rather than the finished form of fully driverless transport.

What challenges will autonomous vehicles create for motor insurers?

The insurance industry will need to adapt to a different combination of risks.

An accident may no longer be explained primarily by the actions of the person sitting behind the wheel.

Instead, insurers may need to investigate the vehicle, its software, the organisations responsible for operating it and the circumstances in which the automated system was being used.

This could affect everything from claims investigations to repair processes and premium calculations.

1. Access to vehicle data

Imagine two apparently identical cars involved in separate collisions.

One was being driven manually. The other had its authorised self-driving system engaged.

That distinction could be central to establishing responsibility.

Insurers may need access to detailed vehicle information to answer questions such as:

• Was the automated system active at the moment of impact?

• Was the vehicle operating on an authorised road and within the permitted speed and weather conditions?

• Did the system ask a human to resume control?

• If it did, was sufficient warning provided?

• Had the required software and security updates been installed?

• Was there a mechanical fault, damaged sensor or cybersecurity incident?

Without reliable information, apparently straightforward claims could become complicated disputes involving insurers, manufacturers, software developers and operators.

However, access to vehicle data must also respect privacy and protect commercially sensitive information.

Clear rules around who can access information, what it can be used for and how it is protected will become increasingly important.

2. More expensive and complicated repairs

Modern vehicles already contain cameras, radar and sensors that can make even minor collision repairs more involved.

Autonomous vehicles could take this further.

A damaged bumper might contain equipment responsible for detecting pedestrians or other vehicles.

Repairing the exterior panel may not be enough.

The sensors and associated systems may need specialist inspection, calibration and testing to ensure they're working correctly.

A vehicle that looks perfectly repaired must still be capable of interpreting the road safely.

These requirements could affect repair costs, repair times and the availability of suitably equipped repairers.

For insurers, this creates an important distinction between how frequently accidents occur and how much each accident costs.

3. A different approach to insurance pricing

Traditional motor insurance uses information such as a driver's age, experience, claims history, vehicle and location to help assess risk.

For a vehicle operating fully autonomously, some of those factors may become less relevant while the automated system is driving.

Instead, insurers may increasingly need to consider factors such as:

• The safety record of the automated driving system.

• The roads and conditions in which it is authorised to operate.

• Software and update history.

• Vehicle maintenance and repair arrangements.

• The organisations responsible for the automated technology.

Two vehicles that look almost identical could present different risks because of their software, sensors or authorised driving capabilities.

The information required to arrange insurance may therefore become more detailed.

For brokers such as Performance Direct, ensuring that insurers receive accurate information about a vehicle's specification and use will remain important.

4. Software and cybersecurity risks

Conventional vehicles can suffer mechanical and electrical failures.

Autonomous vehicles introduce additional concerns around software reliability, digital security and the possibility of a fault affecting multiple vehicles using the same system.

A defect in one software platform could potentially affect a large number of vehicles.

Cybersecurity incidents might also create risks involving vehicle control, operational disruption and access to sensitive information.

Insurers will need to understand how these exposures are managed and how responsibility is allocated when incidents occur.

Vehicle manufacturers and software developers will also need to demonstrate that systems remain safe and secure throughout their operating lives.

5. The difference between assisted and autonomous driving

This is particularly important for motorists.

Many modern cars already offer features such as adaptive cruise control, automatic emergency braking and lane-keeping assistance.

However, the presence of these systems doesn't mean a vehicle is legally authorised to drive itself.

Most driver-assistance systems require the driver to remain attentive and responsible throughout the journey.

An authorised self-driving feature operates under a different legal framework and within specified conditions.

Confusing the two could create dangerous situations.

Drivers should understand the capabilities and limitations of the technology fitted to their vehicle and follow the manufacturer's instructions.

Calling a car self-driving when it still requires constant human supervision could also create serious questions during an insurance claim.

Will driverless cars make insurance cheaper?

It's a question many motorists will understandably want answered.

Well... possibly.

But it would be unwise to assume that autonomous driving will automatically result in cheaper premiums.

If driverless vehicles eventually cause fewer accidents, that could reduce some insurance costs.

However, the frequency of claims is only part of the equation.

Autonomous vehicles can contain expensive technology that requires specialist repairs and recalibration.

Even relatively minor bodywork damage could result in a substantial repair bill.

A defect affecting a widely used software system could also create risks involving multiple vehicles at once.

Insurers will need to balance any reduction in collisions against repair costs, claim severity and new technology-related risks.

Premiums will ultimately depend on real-world evidence.

Insurers will want to understand how frequently particular systems are involved in accidents, how severe those accidents are and how much the vehicles cost to repair.

Until sufficient comparable claims information becomes available, it's too early to establish how autonomous driving will affect premiums across the wider motor insurance market.

The Road Ahead: What Does Autonomous Driving Mean for Motor Insurance?

The arrival of supervised autonomous Uber rides in London marks a significant development in how vehicles are operated.

However, it also demonstrates why the transition to fully driverless transport is likely to be gradual.

For now, a licensed driver remains ready to take control. The wider legal framework is still being implemented, and questions surrounding vehicle data, repairs, cybersecurity and responsibility will continue to develop.

For insurers and brokers, the focus may increasingly shift from assessing how a person drives to understanding how an entire transport system operates.

That includes the vehicle, its software, the organisation responsible for automated driving and the operator keeping it on the road.

Britain's roads will provide a demanding environment in which to establish how well the technology performs.

If autonomous vehicles can deliver improvements in safety and accessibility, they could change how people travel.

But one important principle remains.

Whatever is doing the driving, people involved in an accident need to know who is responsible and how they can obtain compensation.

Cars Are Changing. Make Sure Your Insurance Meets Your Needs.

From everyday motoring to performance, modified and specialist vehicles, having suitable insurance remains important. Explore Performance Direct's car insurance options and get a quote today.

Frequently Asked Questions About Driverless Cars and Insurance

Are London's new Uber taxis completely driverless?

No.

The Uber and Wayve service launched on 3 September 2026 uses supervised autonomous vehicles.

A trained, TfL-licensed private hire driver remains in the driving seat throughout each journey and can intervene when necessary.

Passengers may be matched with an autonomous vehicle through the Uber app and can choose a conventional ride instead.

Can I refuse an autonomous Uber?

Yes.

Uber states that passengers matched with a Wayve vehicle can choose to accept the autonomous ride or switch to a conventional Uber before the vehicle arrives.

There is no additional charge for the autonomous journey compared with the equivalent booking.

Who pays if a driverless car causes an accident?

The arrangements depend on the vehicle's legal status, whether an authorised automated driving feature was operating and the circumstances of the accident.

The Automated and Electric Vehicles Act 2018 provides a framework under which the insurer of a listed automated vehicle can be liable in the first instance for an accident caused while it is driving itself, subject to statutory requirements.

The insurer may subsequently seek to recover costs from another responsible organisation.

Is a car with adaptive cruise control considered self-driving?

Not necessarily.

Adaptive cruise control and lane-keeping assistance are driver-assistance features.

They generally require the driver to remain attentive and responsible for controlling the vehicle.

An authorised self-driving system operates under a different legal framework and within specified conditions.

Motorists should always understand the capabilities and limitations of their vehicle's technology.

Are autonomous cars safer than conventional vehicles?

There isn't a universal answer.

Performance varies between systems, locations and operating conditions.

Automated technology could potentially reduce collisions associated with certain forms of human error, but it also introduces different risks involving software, sensors and cybersecurity.

Reliable, comparable evidence from real-world operation will be needed to understand how particular systems perform.

Will autonomous cars make motor insurance cheaper?

Possibly, but lower premiums aren't guaranteed.

Fewer collisions could reduce some costs, while expensive sensors, specialist repairs and technology-related risks could increase others.

Insurers will need sufficient claims evidence before determining how automated driving affects premiums for different vehicles and circumstances.

When will fully driverless taxis become available in the UK?

London already has supervised autonomous taxi journeys, but these still have a trained driver behind the wheel.

The Government is working towards implementing the wider Automated Vehicles Act 2024 framework in the second half of 2027.

That framework will establish the rules governing authorised vehicles that can operate without a user in charge.

The timing and availability of individual commercial services will depend on regulatory approval and operators' deployment plans.

Will I still need car insurance if my vehicle drives itself?

The development of autonomous driving doesn't remove the need for appropriate insurance.

The legal framework establishes arrangements for insurer liability and assigns responsibilities to organisations involved in authorised automated driving.

The insurance requirements that apply will depend on the vehicle, its use and the relevant legislation.

Motorists should not assume that having an automated driving feature changes their existing insurance obligations.