If the crime-scene breach was dramatic, the Elder Street saga in Shoreditch was absurdly persistent. Between May and June 2026, Waymo’s mapping vehicles repeatedly turned into a dead-end street in the early morning hours—then struggled to reverse out. Residents described the noise as an “awful siren” and a “ridiculous blend of reversing noises and siren sounds” blaring around 4 a.m. .
Footage showed a white Jaguar SUV mounting the curb at 4:15 a.m. while trying to perform a multi-point turn on the narrow street . Even after the problem was publicly reported, locals said at least one vehicle kept returning
. Waymo apologized for the disturbance and said it was working on a fix
.
The deeper worry is systemic. Autonomous vehicles operate on highly detailed maps, and a dead-end misclassification can send a car into a loop that a human would avoid after one mistake. The fact that multiple vehicles made the identical error over weeks suggests a mapping or routing weakness that a simple sensor tweak might not solve.
In May 2026, a dashboard camera captured a Waymo robotaxi entering a busy Dallas intersection against a red light and navigating through moving traffic . No one was injured, but the video went viral.
Waymo responded that the traffic signal was “heavily dimmed” when viewed from the right-turn lane, implying a perception challenge rather than a planning failure . Researchers weren’t convinced; one told CBS News that the tech still needs work
.
Around the same time, a Dallas resident told CBS News she watched a Waymo fail to yield for a firetruck, then continue driving at roughly 20 mph through a school zone—even though the zone wasn’t active . It’s the kind of layered judgment call (emergency-vehicle priority plus school-zone rules) that exposes gaps in autonomous systems’ ability to interpret situational context. CNN’s broader investigation found that the absence of a human operator was introducing “new risks” that regulators are increasingly concerned about as Waymo attempts to scale
.
The most consequential Texas failures involved school buses. Starting in late 2025, surveillance cameras on Austin Independent School District buses caught Waymo vehicles illegally passing stopped buses with their stop arms extended and red lights flashing—at least 19 times since the school year began . Austin ISD police said the vehicles were cited 20 times and that Waymo initially refused to halt operations until the issue was fixed
.
The National Highway Traffic Safety Administration opened an investigation in October 2025 . In December, Waymo voluntarily recalled the software, acknowledging a glitch that caused its vehicles to bypass stopped school buses
. Waymo Chief Safety Officer Mauricio Peña said no collisions occurred, but the optics of schoolchildren being put at risk—even statistically—were damaging
.
On May 22, 2026, Waymo suspended its fully autonomous robotaxi service across Dallas, Austin, Houston, and San Antonio. The trigger: heavy rains and flash flooding, with online videos showing Waymo vehicles stopped in floodwaters in Atlanta after a rainstorm .
The company called the move a “proactive safety measure” and cited severe weather across Texas . No injuries were reported, but the suspension underscored how autonomous fleets remain fragile in bad weather—a well-known Achilles’ heel of lidar-and-camera perception stacks.
Waymo’s playbook across these incidents has been consistent: acknowledge, apologize, patch, and—when necessary—pause. The company issued formal apologies for the London noise disturbances . It suspended the safety driver involved in the crime-scene breach
. It voluntarily recalled software for the school-bus failures
and, earlier, recalled 1,212 vehicles for a risk of hitting fences, gates, and chains
.
Yet the pattern raises a strategic question: is Waymo fixing edge cases one by one, or are these symptoms of a deeper “common sense” deficit? A report on the London tests noted that even though Waymo data shows 12 times fewer injury-causing crashes than human drivers, the incidents “expose ‘common sense’ AI flaws amid safety concerns” .
It’s worth noting that the NHTSA closed a 14-month probe into Waymo in July 2025 without finding systemic issues . And in the grand statistical picture, autonomous vehicles do tend to be safer than human drivers overall
. But the London-Dallas cluster shows that public trust is built not on fleet-wide averages but on vivid, viral moments where the tech looks dangerously out of its depth.
Comments
0 comments