Waymo is actively trying to stop children from riding alone in its driverless cars in California, where state rules flatly prohibit it, but San Francisco parents have been quietly ignoring that policy for months, creating a regulatory, liability, and trust crisis that reveals how far robotaxi operations are from scaling cleanly into everyday family life.
In January 2026, the New York Post and SFist both reported the same phenomenon. Affluent San Francisco families were booking Waymo rides for their children, solo, using the parent's account, to get them to school, sports practice, and social activities without having to drive. The cars kept picking up the kids. The kids kept arriving safely. Parents kept booking. By March, the California Gig Workers Union had filed a formal complaint with the California Public Utilities Commission, alleging that Waymo was violating the conditions of its operating permit by transporting unaccompanied minors. The CPUC's 2020 rules governing autonomous vehicle cab services state plainly: no one under 18 may travel unaccompanied, regardless of whoever chartered the vehicle. Waymo knew about the gap, had applied to change the rule, and been rejected. The cars kept running.
The regulatory situation is genuinely tangled. Waymo's own terms of service prohibit minors from riding alone in California. The company's website says it. The CPUC permit says it. It is also unenforceable in any practical sense with the current technology stack. Waymo acknowledged this directly in its own CPUC filings: "It is not feasible for Waymo to categorically check and reliably determine the age of every rider." A parent creates an account, attests they are over 18, and can then book rides for whoever gets in the car. There is no biometric check at the vehicle door, no mandatory age verification step before each ride, and no real-time system that can distinguish a 35-year-old from a 13-year-old booking the same trip through a verified adult account. The enforcement gap is not theoretical. It is operational.
The comparison to Phoenix makes the gap look even more deliberate. Waymo launched a formal teen account program for riders aged 14 to 17 in the Phoenix metro area in mid-2025. Parents link the teen's account to their own, grant permission, and the teen can hail rides independently within approved zones. The cars have enhanced monitoring features active during teen rides. Remote support operators are on standby. There are real-time trip notifications for the parent. Waymo has shown it can build an architecture for younger riders. The Phoenix rollout is proof of concept. The reason it does not exist in San Francisco or Los Angeles is regulatory, not technical: California's CPUC rules prohibit it, and Waymo's application to change that was rejected.
The California Gig Workers Union's complaint is transparently motivated in part by competitive interest: its members drive for Uber and Lyft, which are direct rivals of Waymo. That does not make the underlying safety concern invalid. The union's spokesperson, an Antioch-based Lyft driver, framed it accurately: "If something goes wrong, there's no one there to protect them, comfort them, or make sure they're safe." Waymo's remote support model means a human operator can monitor a trip and communicate via the vehicle's interior screen, but they cannot physically intervene. If a child has a medical emergency, is frightened, or encounters a situation the vehicle cannot navigate, the response options are limited to verbal guidance and dispatching emergency services. That is not nothing, but it is structurally different from having an adult in the car.
The privacy dimension runs directly against the safety solution. The most obvious technical fix for age verification is interior camera monitoring and computer vision classification of passenger age at the moment of pickup. Waymo's vehicles already have interior cameras that serve safety and support functions. Activating them more aggressively for age screening would immediately generate a different kind of controversy. The same parents who happily put their children in unaccompanied Waymos would likely object to a company using AI-driven surveillance to classify their family's biometric data at every pickup. Privacy advocates would object regardless of whether parents consented. Any system Waymo builds to solve the minor detection problem will be audited for what else it detects, stores, and shares.
The market demand underlying all of this is real and growing. Transport to school and activities is one of the largest unmet pain points in suburban and urban family life. Uber and Lyft have tried teen programs with human drivers in some markets. Waymo's autonomous model theoretically offers a safer environment precisely because there is no adult stranger in the front seat with access to the child. The Phoenix teen program demonstrated meaningful consumer demand. Families who have used it report high satisfaction. The regulatory barrier in California is not a reflection of genuine safety consensus, it is the product of 2020 rulemaking that did not anticipate how the consumer appetite for this service would develop.
What Wired's reporting and the CPUC complaint together reveal is that the robotaxi industry has reached a stage where the technology is ahead of the regulatory framework, and where consumer behaviour is ahead of both. People are using autonomous vehicles to solve real daily problems in ways the product team did not officially sanction. That is usually a sign of genuine product-market fit. It is also a sign that the liability structure, the enforcement mechanisms, and the public trust frameworks have not caught up. Waymo's challenge is not primarily technical. It is figuring out how to serve the family market, which is clearly asking for its product, without accumulating the regulatory and legal exposure that comes from doing so in jurisdictions that have not yet decided whether they want to allow it.
","excerpt":"Waymo is cracking down on solo child riders in California after a formal CPUC complaint, while its Phoenix teen account program demonstrates the market demand is real.
Also read: A Developer Ditched Cloud AI for a Local Model on an RTX 6000 Pro and the Results Are Hard to Argue With • Unsloth found and fixed a bug in Mistral Medium 3.5 and the story reveals how much of the open-weight model race is actually a tooling race • Viral prompts for fake old photos reveal how synthetic nostalgia is becoming one of AI image generation's most persuasive and least discussed capabilities
