Mind the Weymo! The Robotaxi Nobody Asked For, Coming To a Roundabout Near You

Silicon Valley's pride and joy is about to meet its match on the Old Kent Road.

There is a particular kind of announcement that arrives from Silicon Valley with the energy of someone who has just discovered something the rest of the world has been doing quietly for decades. In October 2025, Waymo, the self-driving car subsidiary of Alphabet, which is itself a subsidiary of the collective anxiety of anyone who remembers when Google just did search, declared that it would bring its fully autonomous ride-hailing service to London, making our sodden, chaotic, magnificently unsuitable capital its first venture outside the United States.

The press release was the usual thing: transformative, revolutionary, a new era of mobility. By December, Waymo vehicles had already been photographed gliding across the Abbey Road zebra crossing, presumably because nothing says "we understand London" quite like a reference that peaked in 1969. The company posted it on X with Union Jack emojis. Demis Hassabis, CEO of Google DeepMind, retweeted it with barely concealed glee. Somewhere in Whitechapel, a man who has spent eleven years doing the Knowledge looked at his phone and poured himself something stronger than tea.

What we are being offered, in essence, is a robotaxi. A very expensive, very American robotaxi, staffed, and we will return to this word "staffed", by sensors, algorithms, and a Level 4 autonomous system that has been extensively tested on the wide, rational, grid-patterned streets of Phoenix, Arizona. London, for those who need reminding, is not Phoenix, Arizona. London is a city whose road layout was determined largely by the hooves of medieval livestock, whose cyclists operate under a philosophy best described as "theological," and whose pedestrians treat a green man as a suggestion and a red one as a personal affront. The Waymo vehicle that handles Chandler Boulevard with sedate competence is about to meet the Elephant and Castle roundabout at rush hour. We genuinely cannot wait.

Actually, How Safe Is It?

Waymo would very much like to talk to you about its safety record. Understandably so, because on paper it is impressive. By mid-2025, the company had completed over 127 million fully autonomous miles. Peer-reviewed studies found its vehicles involved in 91% fewer serious-injury crashes and 80% fewer injury-causing incidents than human drivers on equivalent roads. Swiss Re, an insurer, and therefore a body with a financial incentive to be accurate about risk, analysed 25 million Waymo miles and found 92% fewer bodily injury claims than comparable human-driven vehicles. These are not trivial numbers. They are the kind of numbers that should, in an ideal world, settle the argument.

This is not, alas, an ideal world. It is 2026, and the first weeks of that year have been what the company's PR department might gently describe as a learning period.

On 23rd January, a Waymo robotaxi struck a child outside Grant Elementary School in Santa Monica during morning drop-off. The child ran from behind a double-parked SUV. The vehicle was travelling at 17 mph, within the posted limit, and there was no safety operator inside. Waymo's response, delivered with the emotional intelligence of a man reading terms and conditions at a funeral, was to note that a peer-reviewed model suggested a fully attentive human driver would have hit the child at roughly 14 mph, nearly double the impact speed. This is almost certainly true. It is also, as communications go, roughly equivalent to responding to a restaurant food poisoning complaint by observing that home cooking kills more people statistically. Accurate, tone-deaf, and rather missing the point that a child was struck by one of your cars outside a primary school.

It did not stop there. The National Transportation Safety Board opened a separate investigation after it emerged that Waymo robotaxis had illegally passed stopped school buses, stop signs extended, lights flashing, at least 19 times in Austin, Texas. The company subsequently issued a recall of over 3,000 vehicles to patch the software responsible. A recall. Not a tweak, not an update quietly pushed overnight: a recall, the kind of word that belongs to the vocabulary of faulty toasters and exploding airbags. Federal investigators from the National Highway Traffic Safety Administration opened their own parallel inquiry. By the end of January, Waymo had two simultaneous federal probes and a story about hitting a child outside a school. The Union Jack emojis felt a long way away.

To be fair, and the article will be fair, even if Waymo's timing has not been, the aggregate safety data remains genuinely significant. The school zone incidents are serious, but they represent a specific failure mode in a specific context, not evidence that the whole enterprise is a charade. The more substantive critique is that Waymo's statistics are drawn from every mile it has ever driven, including empty streets at three in the morning and long arterial roads with clear sightlines. They do not break out performance in the particular, irreducible chaos of the school run: double-parked 4x4s, children who have not yet grasped the concept of the kerb, and the kind of unpredictability that has defeated more than one human driver too. That is precisely the environment regulators are now examining. It is also, for what it's worth, the environment London has in abundance, morning and afternoon, five days a week, forty weeks a year.

Not Quite Driverless (Please Read the Small Print)

Here is where it gets interesting. Waymo has built its entire public identity around the concept of the car that drives itself, no human, no intervention, pure algorithmic competence steering you home. It is a clean story. It is also, as a US Senate hearing in February 2026 established, not entirely the whole story.

Pressed by senators on what happens when a Waymo vehicle encounters a situation it cannot resolve, the company's chief safety officer Mauricio Peña confirmed that the vehicles can call upon remote "fleet response agents", workers who monitor feeds, review sensor data, and provide guidance when the software requests it. Some of these agents are based in the United States. Others, Peña acknowledged, are based in the Philippines. When asked precisely how many, he said he did not have the breakdown to hand. The senators' frustration was palpable. The public's right to mild bewilderment seems equally justified.

Waymo's framing of this arrangement is careful. The remote agents, it says, do not drive the cars. They provide high-level navigational guidance, essentially, a very expensive version of asking someone to look something up, and the vehicle retains control of all physical driving tasks. The human cannot grab the wheel, because there is no wheel to grab from Manila. Technically, the car remains autonomous. In the rather larger sense of the word, a technology marketed as the elimination of the driver has discovered that you cannot quite eliminate the driver; you can only make the driver invisible, ineligible for employment protections, and based somewhere with lower wage expectations.

This is a story that will be familiar to anyone who has spent time watching how Silicon Valley actually operates beneath its own press releases. The gig economy promised flexibility and delivered precarity. Artificial intelligence promised automation and quietly hired an army of underpaid data labellers in the Global South to do the bits the algorithm couldn't manage. Now autonomous vehicles promise driverlessness and turn out to rely on a distributed workforce of remote monitors who are essential enough to be employed but apparently not prominent enough to mention in the brochure. Senator Ed Markey put it with admirable plainness: "It's one thing when a taxi is replaced by an Uber or a Lyft. It's another thing when the jobs just go completely overseas." London's cabbies, who have watched their industry hollowed out over a decade, may be forgiven for finding this particular detail less than reassuring.

The Part Where We've Definitely Seen This Before

Those of a certain age and address will remember the summer of 2012. Uber arrived in London that July, timed neatly for the Olympic Games and trailing the kind of rhetoric that would become very familiar: disruption, democratisation, a better deal for everyone. Travis Kalanick called London "the gold standard" of markets. He said this with a straight face.

What followed was a decade of low-grade warfare between a Silicon Valley platform and one of the world's oldest urban transport ecosystems. Black cab drivers, who had spent between three and four years and somewhere north of £20,000 acquiring the Knowledge, memorising 25,000 streets and every point of interest within six miles of Charing Cross, found themselves undercut by drivers following a phone. Uber classified those drivers as independent contractors, denied them employment rights, and initially deployed software specifically designed to show regulators a version of the app that wasn't the real version of the app. Transport for London stripped Uber of its operating licence in 2017, citing a "pattern of failures." Then again in 2019. Each time, Uber appealed, made promises, and was eventually permitted to continue. It currently serves millions of Londoners daily. The black cab fleet has contracted from roughly 22,800 vehicles in 2014 to fewer than 15,000 today. The disruption was very much democratised. The people disrupted were the drivers.

Waymo's arrival rhymes with this story without quite repeating it. The company is not doing anything illegal. It has arrived with the backing of the Automated Vehicles Act 2024, the explicit support of the Secretary of State for Transport, and a safety record it can discuss at length using peer-reviewed studies. It is not running Greyball. It is not classifying anyone as an independent contractor. It is, in one sense, more honest than Uber ever was about what it intends to do: it intends to remove the driver entirely. No exploitation of the workforce because there is no workforce. Just software, sensors, and, quietly, in the background, those fleet response agents in the Philippines, whose employment conditions remain, at the time of writing, undisclosed.

Steve McNamara, General Secretary of the Licensed Taxi Drivers' Association, called Waymo's initial press release "incredibly misleading" and expressed confidence that his members would survive this as they have survived everything else. He also mentioned, with what one hopes was deliberate comedy, that he owns a robot lawnmower. He is not wrong that cabbies have endured before. He would be unwise to treat that endurance as a guarantee. The market does not evict its incumbents cleanly, it layers over them, compresses them, and leaves them serving a clientele that either cannot or prefers not to use the new thing. There are still black cabs. There are also far fewer of them, and the people inside tend to be either corporate account holders or tourists who find the whole experience charming, which is a slightly different customer base from the one that sustained the trade for a century.

The Tesla-Shaped Cloud Over Everything

Before Waymo can have the conversation it wants to have with the British public, sensible, evidence-based, here are the peer-reviewed statistics, it must first deal with a conversation it did not start and cannot entirely escape. That conversation was started by Elon Musk, who has spent the better part of a decade insisting that Tesla vehicles are perpetually six months away from full autonomy, that Full Self-Driving is exactly what the name suggests, and that anyone who disagrees is a fool or a Luddite or possibly both.

Full Self-Driving is not full self-driving. It is a driver assistance system, Level 2, in the industry's taxonomy, that requires the driver's continuous attention, has been linked to hundreds of crashes, and carries a name so misleading that multiple regulators in multiple countries have had pointed things to say about it. Tesla sales across Europe fell by extraordinary margins through 2024 and 2025, driven partly by Musk's increasingly conspicuous political activities, a man who spent his evenings on stage with Donald Trump and his weekends doing things in Berlin that older Europeans found historically resonant was, it turned out, not the ideal brand ambassador for the continental market. In some months, European Tesla sales were down 40%.

The damage to public trust in autonomous vehicle technology generally has been real and lasting. A survey of more than 8,000 Americans in August 2025 found that 35% said Full Self-Driving made them less likely to buy a Tesla, a remarkable result for a feature that is meant to be a selling point, while nearly half believed such technology should be illegal. In the United Kingdom, 52% of respondents in a separate survey said they would not trust self-driving cars at all. This is the atmosphere into which Waymo is attempting to launch a business.

Waymo is not Tesla. On a technical level, this comparison is roughly as meaningful as comparing a qualified surgeon to someone who watched a lot of medical dramas and feels confident. Waymo's Level 4 system operates without a driver in a defined geography, has logged hundreds of millions of real-world miles, and has never told anyone that immortality is just around the corner. It has earned its safety statistics slowly and, largely, without catastrophe. But the British public, encountering a driverless car on the streets of Hackney, is not consulting the technical taxonomy of vehicle autonomy. It is thinking about the footage it has seen of Teslas ploughing into stationary vehicles, of Autopilot engaging on motorways, of a chief executive who named the feature and then spent years defending it while his cars crashed into things. Waymo will spend considerable time and money paying for sins it did not commit. That is simply the price of arriving after Tesla.

The Road Ahead, With Potholes

Waymo will begin, sensibly, small. A limited fleet of electric Jaguar I-Pace vehicles, the ones with the sensor towers on the roof that make them look like they are wearing a very elaborate hat, will operate in selected London boroughs, under close regulatory supervision, awaiting full sign-off under the Automated Vehicles Act. The vehicles are already a recognisable sight in parts of Camden, Westminster, and Hackney. The full commercial service, entirely without a safety driver, is not expected before late 2027 at the earliest. London's actual transformation into the setting of a science fiction film remains, for now, on the schedule but not in the diary.

There is a case, and it should be made, that some of this is genuinely good. For people with visual impairments, for those whose disabilities make conventional transport inaccessible, for the elderly person who no longer drives but lives somewhere the 43 bus does not reach, an autonomous vehicle that arrives on request and requires no social navigation is not a luxury. It is a meaningful change. The Royal National Institute of Blind People called Waymo's announcement "the potential dawn of a new era in independent mobility options," and that framing deserves to be taken seriously, not buried under the reasonable scepticism the technology has earned through its behaviour in school zones.

But for the average Londoner, possessed of an Oyster card, a moderate tolerance for the Northern line, and a bone-deep suspicion of anything that arrives promising revolution, the proposition is less clear. London does not have a shortage of ways to get from one place to another. It has a shortage of affordable housing, functioning GP surgeries, and bus routes that haven't been cut, but taxis are not the gap in the market Waymo's press release implies. What it is offering is a premium product dressed in the language of public good, backed by safety statistics that were genuinely impressive until January, dependent on overseas labour it prefers not to discuss, and entering a city that has already watched one technology company transform its transport landscape for better and considerably worse.

The black cab driver who memorised this city before the algorithm existed deserves more than to be dismissed as a charming anachronism. The Filipino remote monitor who guides Waymo's cars through situations the software cannot handle deserves to have their existence acknowledged. And the child outside Grant Elementary School deserves a technology that has fully reckoned with what it means to operate heavy machinery near primary schools, rather than one that responds to her injury with a comparative impact velocity analysis.

Waymo may yet earn London's trust. It will do so not through press releases, not through Union Jack emojis, and certainly not through telling us that hitting a child slowly is better than hitting one quickly. It will do so, if it does so at all, by working: quietly, consistently, without incident, across the narrow streets and deranged roundabouts and bicycle-strewn junctions of this deeply unsuitable city, until the novelty fades and what remains is simply something that functions. London has seen enough transformation promised and partially delivered to know the difference between the two. It will wait and see. It is very good at that.

References

Di Lillo, L., Gode, T., Zhou, X., Atzei, M., Chen, R. and Victor, T. (2024) 'Comparative safety performance of autonomous- and human drivers: a real-world case study of the Waymo Driver', Heliyon, 10(14), e34379. Available at: https://doi.org/10.1016/j.heliyon.2024.e34379 (Accessed: 2 March 2026).

Kusano, K.D., Scanlon, J.M., Chen, Y.H., McMurry, T.L., Gode, T. and Victor, T. (2025) 'Comparison of Waymo Rider-Only crash rates by crash type to human benchmarks at 56.7 million miles', Traffic Injury Prevention, 26(sup1), pp. S8–S20. Available at: https://doi.org/10.1080/15389588.2025.2499887 (Accessed: 2 March 2026).

Tennant, C. and Stilgoe, J., Vucevic, S. and Stares, S. (2024) 'Public anticipations of self-driving vehicles in the UK and US', Mobilities, 20(2), pp. 292–309. Available at: https://doi.org/10.1080/17450101.2024.2325386 (Accessed: 2 March 2026).

Park, I.R., Kim, S. and Moon, J. (2025) 'Why do people resist AI-based autonomous cars? Analyzing the impact of the risk perception paradigm and conditional value on public acceptance of autonomous vehicles', PLOS ONE, 20(2), e0313143. Available at: https://doi.org/10.1371/journal.pone.0313143 (Accessed: 2 March 2026).

National Highway Traffic Safety Administration (2026) Preliminary Evaluation: Waymo Automated Driving System — School Zone Incident, PE26001. Washington, D.C.: NHTSA. Available at: https://static.nhtsa.gov/odi/inv/2026/INOA-PE26001-10005.pdf (Accessed: 2 March 2026).

UK Parliament (2024) Automated Vehicles Act 2024. London: HMSO. Available at: https://www.legislation.gov.uk/ukpga/2024/10/contents (Accessed: 2 March 2026).

Dodds, C. (2026) 'Waymo reveals remote workers in Philippines help guide its driverless cars', Newsweek, 6 February. Available at: https://www.newsweek.com/waymo-reveals-remote-workers-in-philippines-help-guide-its-driverless-cars-11478439 (Accessed: 2 March 2026).

Hogg, R. (2024) 'Uber faces a 9-figure lawsuit that could net London taxi drivers $31,000 each', Fortune, 2 May. Available at: https://fortune.com/europe/2024/05/02/uber-9-figure-lawsuit-london-taxi-black-cab-drivers-31000/ (Accessed: 2 March 2026).

Grenoble, R. (2017) 'Uber Has A Secret Program Called 'Greyball' It Uses To Evade Police', The Huffington Post, 9 March. Available at: https://www.huffingtonpost.co.uk/entry/uber-greyball-software-identify-evade-police_n_58b9dd02e4b0b9989417a1a4 (Accessed: 2 March 2026).

Transport for London (no date) 'Learn the Knowledge of London'. Available at: https://tfl.gov.uk/info-for/taxis-and-private-hire/licensing/learn-the-knowledge-of-london (Accessed: 2 March 2026).

United States Senate Committee on Commerce, Science, and Transportation (2026) 'Hit the Road, Mac: The Future of Self-Driving Cars', 4 February. Testimony of Dr. Mauricio Peña, Chief Safety Officer, Waymo LLC. Available at: https://www.commerce.senate.gov/2026/2/hit-the-road-mac-the-future-of-self-driving-cars (Accessed: 2 March 2026).