Uber and British artificial-intelligence company Wayve have launched supervised autonomous rides in London, giving a limited group of Uber users the chance to travel in vehicles controlled by Wayve’s driving system while a trained private-hire driver remains behind the wheel.
The service began on September 2, 2026. Reuters reported the launch as London’s first robotaxi service and noted that the initial rides include a licensed operator who can monitor the vehicle before any future move to fully driverless operation. Wayve’s public ride page confirms that its autonomous rides are available through Uber and explicitly says a trained driver is onboard under the United Kingdom’s code of practice for automated-vehicle trials.
That human presence is the most important qualification. The vehicles may handle the driving task, but this phase is supervised. It should not be described as a citywide, driverless service. The Verge and Axios also reported that safety drivers remain in the vehicles, independently supporting the distinction between today’s operation and a possible future driverless service.
H2: What launched in London
The initial service uses all-electric Ford Mustang Mach-E vehicles fitted with Wayve’s AI Driver technology. Uber and Wayve say the vehicles use surrounding cameras and radar. A trained and Transport for London-licensed private-hire driver sits in the driver’s seat to oversee the trip, provide support and take control when necessary.
Wayve says riders can indicate a preference for autonomous vehicles in the Uber app. A preference does not guarantee a match because the available fleet is limited. The early service operates within Greater London and excludes airport journeys, according to Wayve’s ride information.
The service follows regulatory and licensing steps announced in August. Uber said Transport for London had granted private-hire vehicle licences to several Wayve-equipped vehicles. The companies described a “triple-lock” arrangement in which the operator, vehicle and driver hold appropriate licences with the same licensing authority. Trips remain subject to the government’s automated-vehicle trialling code.
Calling the vehicles robotaxis is understandable because the automated system performs the driving function. Calling them fully driverless is not accurate at this stage because a licensed human driver is present and responsible for monitoring and intervention.
H2: Why supervised autonomous is the accurate description
Autonomous-vehicle terminology often compresses several different operating models into one word. A demonstration on a closed track, a supervised public-road trial, a commercial service with a safety operator and a driverless service without anyone behind the wheel are not equivalent.
The London launch sits in the supervised category. Customers can receive rides through a normal commercial platform, but the onboard driver remains a safety layer. That changes how the launch should be evaluated.
A supervised deployment can provide useful operational data. It can reveal how the system handles complex junctions, cyclists, buses, pedestrians, roadworks, unusual lane markings and rapidly changing weather. It can also test dispatch, cleaning, charging, customer support and handoffs between the technology provider and ride operator.
However, supervised performance does not automatically predict driverless performance. A safety driver can prevent or reduce the consequences of an error. Until the companies disclose intervention data and regulators authorize removal of the driver, the service remains a step toward driverless operations rather than evidence that the transition is complete.
H2: How riders access the service
Wayve’s public instructions direct users to open the Uber app, enter account settings, select ride preferences and enable the autonomous-vehicle option. Riders who opt in may be matched with a Wayve vehicle when one is available.
The experience should still include clear disclosure before the trip. A rider needs to know that the vehicle uses an automated driving system, that a trained driver is onboard, what the driver may do, how to request help and how trip data is handled.
Riders should also be able to decline the autonomous option without losing access to ordinary transportation. Consent is more meaningful when the choice is clear and reversible.
The limited fleet means the service is best understood as an early operational deployment. It is not yet a universal Uber category available on demand across every London neighbourhood or airport route.
H2: What Wayve’s technology is intended to do
Wayve describes its approach as embodied AI: a driving system trained to interpret its surroundings and generate driving behaviour rather than relying entirely on a rigid, hand-coded rule for every situation. The company promotes an “AV2.0” model intended to generalize across different vehicles and locations.
The London vehicles use cameras and radar. Wayve argues that its approach can adapt to the complexity of real roads without depending on the same combination of high-definition maps and sensors used by some competitors.
Those are company descriptions and strategic claims. The launch does not by itself prove that one sensor or mapping approach is safer, more scalable or less expensive than another. Comparisons require equivalent operating conditions, disclosed safety metrics and enough exposure to rare events.
The practical question is not whether the system uses more or fewer sensors. It is whether the complete vehicle, software, monitoring, maintenance and operational process manages risk consistently within its approved operating conditions.
H2: Why London is an important test
London is a globally visible and operationally demanding market. Roads vary in width and layout. Traffic includes buses, bicycles, motorcycles, pedestrians, delivery vehicles and traditional taxis. Temporary works and changing conditions are common.
A deployment here can test whether an autonomous system transfers beyond the road environments where it was developed. It also tests the business model: whether a technology supplier, vehicle operator, regulator and ride-hailing platform can coordinate safely and reliably.
Reuters described London as the second European city in which Uber offers robotaxi rides after Zagreb. Competitors are also preparing autonomous services for the city, increasing pressure to demonstrate credible performance rather than make broad claims.
The competitive importance should not overshadow public safety. A first-mover advantage matters to companies, but regulators and riders need evidence about collisions, interventions, rule compliance, accessibility, privacy and incident response.
H2: The DRIVER framework for judging the launch
The DRIVER framework separates a real deployment from the larger promise attached to it.
Deployment scope: Identify the number and type of vehicles, geographic area, operating hours, weather limits, airport exclusions and who can receive rides.
Responsible human: Confirm whether a trained driver, remote operator or no human is responsible for supervision. State when and how intervention occurs.
Independent oversight: Look for licensing, regulatory conditions, incident-reporting rules and public authority review. Company testing is useful but not a substitute for oversight.
Verified performance: Ask for exposure measures, collisions, disengagements or interventions, service completion, rule compliance and customer-support outcomes. A raw event count needs a denominator and operating context.
Emergency response: Examine procedures for a stopped vehicle, collision, medical emergency, communications failure, software fault or blocked road. The rider must know how to obtain assistance.
Rider choice and data: Check whether riders knowingly opt in, can decline, receive accessible information and understand what video, location and trip data is collected.
The framework avoids two common mistakes: dismissing the launch because a human remains onboard, or treating a supervised deployment as proof that the system is ready to operate everywhere without a driver.
H2: What is confirmed and what remains unknown
Confirmed facts include the start of supervised rides in London, use of Wayve-equipped Ford Mustang Mach-E vehicles, availability through Uber, the presence of trained private-hire drivers, Greater London operation and the exclusion of airport trips during the trial phase.
The public material does not establish several broader claims. It does not prove a date for removing the driver. It does not demonstrate citywide availability, publish a complete operating-domain specification or provide enough independent performance data to compare the service fairly with other autonomous systems.
Reuters reported that fully driverless operation is intended at some point in the future, but that is a future objective. Regulatory approval, system performance and operational readiness will determine whether and when it happens.
The companies also have not publicly established through this launch alone that supervised autonomous rides will cost less, reduce congestion, improve accessibility or replace a material share of conventional trips. Those outcomes depend on deployment scale and policy choices.
H2: What riders, regulators and competitors should watch
First, watch the boundary of service. Expansion in vehicles, hours, geography and weather conditions can show growing operational confidence, but each expansion changes the risk profile.
Second, watch intervention and incident reporting. Useful disclosure should explain what counts as an intervention, provide exposure data and distinguish routine caution from safety-critical takeovers.
Third, watch the route from supervision to driverless operation. Removing the onboard driver is not simply a software update. It changes fallback responsibility, passenger support, emergency coordination, insurance and regulatory requirements.
Fourth, watch rider behaviour. Opt-in rates, repeat use, complaints and accessibility feedback will show whether customers understand and accept the experience.
Finally, watch competitive claims. London may become a comparison point for different autonomous-driving architectures. Fair comparisons must use similar road complexity, operating hours, supervision levels and reporting standards.
H2: Bottom line
The Uber-Wayve launch is a significant step because ordinary riders can now receive supervised autonomous trips through a mainstream ride-hailing platform in London. It moves the project from sign-ups and licensing into real service.
Its limits are equally important. A trained driver remains onboard, the fleet and service area are constrained, and fully driverless operation is a future possibility rather than a current fact. The most credible assessment will follow the evidence: what conditions the system handles, how often humans intervene, how incidents are reported and how regulators respond as the deployment grows.
ORIGINAL ELEMENT
DRIVER evidence table
Dimension | Evidence to request | Avoid assuming
Deployment scope | Fleet, area, hours and operating limits | Citywide availability
Responsible human | Driver duties and intervention process | No human supervision
Independent oversight | Licences, code and reporting duties | Company approval equals regulatory approval
Verified performance | Exposure and intervention metrics | A demonstration proves safety
Emergency response | Passenger and roadside procedures | Connectivity never fails
Rider choice and data | Opt-in, accessibility and privacy notices | Riders fully understand by default
FAQ
H3: Are Uber and Wayve’s London robotaxis fully driverless?
No. A trained, licensed private-hire driver remains onboard to supervise and take control if needed.
H3: Where are the rides available?
Wayve says the trial operates in Greater London and excludes airport trips. Availability depends on a limited fleet and matching.
H3: How can a rider request one?
Uber users can enable an autonomous-vehicle preference in account settings. Opting in does not guarantee a match.
H3: What vehicles are being used?
The initial service uses electric Ford Mustang Mach-E vehicles equipped with Wayve’s AI Driver, cameras and radar.
H3: When will the service remove safety drivers?
No confirmed public date is established in the sources reviewed. Any change depends on regulatory approval and demonstrated readiness.
