Uber and British AI firm Wayve launched London's first robotaxi service this week, becoming the first company to offer autonomous rides to paying passengers in the UK.
The rollout starts small, with 15 licensed vehicles, each still carrying a human safety driver ready to take control if needed, and Uber has said it does not intend to remove that safety driver imminently.
Uber said more than 100,000 London customers had indicated through the app that they would choose a self-driving taxi if available, though the company's own head of autonomous mobility described the current phase as the first step toward eventual "driver out" technology rather than a race to remove drivers immediately.
Waymo, owned by Google's parent Alphabet, has separately been running mapping trials with safety drivers on London's streets for months and has said it plans its own driverless public service in the city, while Baidu's Apollo Go technology is expected to launch in London through Lyft. London becomes only the second European city, after Zagreb, to host a public robotaxi service.
Responding to the launch, David McMullen, senior organiser of trade union GMB, said the union believes real caution is needed given how many people currently earn a living driving.
"With hundreds of thousands of people employed to drive every day, we need to be really careful with the roll out of driverless cars unless we are prepared to see unprecedented levels of social and economic disruption," McMullen said. He said GMB has called on Transport for London and the government to urgently take responsibility for developing a plan that protects drivers and holds operators accountable for the disruption these services will ultimately cause.
Until such a plan exists, McMullen said "it is premature to move to 'driverless' services," and that the implications need considering before it's too late.
GMB has raised similar concerns previously, warning that more than 300,000 driving jobs could be at risk from the wider rollout of autonomous vehicles, and calling on government to legislate protections for taxi and private hire drivers against job losses and reduced earnings. London Assembly's Transport Committee has separately said it would not expect TfL to grant consent for full autonomous operation at this stage, citing unresolved questions about passenger safeguarding.
The launch is the first practical test of a question Insurance Business UK has covered in detail: who actually carries liability once a human is no longer driving. Under the UK's Automated Vehicles Act, two new legal entities are created to answer that question, the Automated Self-Driving Entity, typically the vehicle manufacturer or technology provider, and the No-User-In-Charge operator, which could be a mobility service provider such as Uber or Waymo.
As Ben Gardner, a partner at law firm Shoosmiths, has put it, the Act essentially creates those two new legal actors to which liability will now attach, replacing the traditional post-crash process of courts untangling blame between drivers.
That framework offers clarity on paper, but its practical insurance impact remains genuinely unresolved. Insurance Business UK has previously reported that liability is migrating from the driver to the technology stack as underwriting models evolve, with Simon Smith, claims director at Carpenters Group, noting that London's busy urban roads, particularly areas with high pedestrian and cyclist activity, create additional uncertainty for autonomous systems that may make some insurers more cautious.
Smith pointed to the likely need for a layered insurance offering combining traditional motor liability, still relevant when a robotaxi is struck by a third party, alongside newer coverage forms such as technology errors and omissions insurance already used in the US market, designed to respond to software failures rather than driver error.
The current phase, with a licensed safety driver still present, sits in a genuine grey area for insurers: the vehicle is designed and marketed as autonomous, but a human remains legally and physically capable of intervening, which is exactly what happened during at least one ride a BBC journalist took at launch, when the safety driver intervened mid-journey.
That hybrid state matters because it determines whether an incident is treated as conventional motor liability resting with the safety driver, or as a technology failure attaching to Wayve or Uber under the AV Act's new entities, a distinction that has direct consequences for which insurance policy, and which insurer, actually pays a claim.
Fitch Ratings has previously warned that insurers who fail to adapt to this shift risk fading away as risk coverage moves from personal motor policies toward commercial product liability, with data gathered from autonomous vehicles playing a growing role in how underwriters price risk as the traditional personal motor risk pool shrinks over time.
London's genuinely difficult driving conditions, narrow, centuries-old streets, dense pedestrian and cyclist activity, and famously unpredictable weather, make this launch a more demanding real-world test of autonomous vehicle technology than the wide, grid-planned streets of Phoenix or San Francisco where robotaxis have operated for longer.
For insurers, that difficulty is precisely the point: London is where the AV Act's liability framework, and the layered insurance products built around it, will actually be stress-tested against genuinely difficult driving conditions rather than remaining a matter of regulatory theory.
GMB's warning about social and economic disruption for drivers and the insurance industry's own unresolved questions about who pays when something goes wrong are, in a sense, two sides of the same transition: one about the human cost of removing drivers, the other about whether the framework built to replace their liability actually works once tested at scale.
