Federal safety regulators are once again turning their attention to Waymo’s driverless fleet, this time to determine whether a software remedy the company rolled out last year genuinely eliminated a dangerous behavior: robotaxis passing school buses that have their stop arms extended and red lights flashing. The National Highway Traffic Safety Administration (NHTSA) opened the review on November 6, 2026, and the central question is not whether Waymo tried to fix the problem, but whether the fix held.
- Scope of the review: NHTSA’s recall query covers roughly 2,895 Waymo vehicles running the company’s fifth-generation automated driving system.
- The timing problem: Regulators say most reported incidents occurred after November 17, 2025, the date Waymo says its corrective software had reached every affected vehicle.
- Not a penalty: A recall query is an investigation into the adequacy of a remedy, not a finding that Waymo is liable or that the update failed.
What Regulators Are Actually Asking
The inquiry, identified as RQ26006 in a document dated November 6, 2026, follows an earlier NHTSA investigation opened in October 2025 after reports surfaced of robotaxis ignoring school bus signals. Waymo subsequently filed a voluntary recall on December 10, 2025, covering 3,067 vehicles, and stated that the updated software had already been installed between November 5 and November 17 of that year.
According to Waymo’s recall filing, the underlying defect was a prioritization error: the system could decide that remaining stopped for a bus was obstructing other traffic and move off, even while the bus’s warning lights were still flashing or its stop arm remained extended. The remedy was designed to prevent the software from placing the movement of other vehicles above staying put for a school bus displaying its signals.
NHTSA now wants to know why reports kept arriving after that remedy date. The agency is examining both whether the original recall adequately addressed the defect and what caused the incidents that followed it. The review covers 2,895 vehicles equipped with the fifth-generation driving system.
Documented Incidents and a Remote-Assistance Wrinkle
The case history is unusually detailed. The Austin Independent School District documented 19 incidents involving Waymo vehicles passing stopped school buses during the 2025 school year, according to correspondence submitted to federal regulators, and the district went as far as asking Waymo to suspend operations during common pickup and drop-off hours.
In Atlanta, a Waymo vehicle reportedly stopped initially for a bus unloading children, then drove around its front and continued down the road while the red lights flashed and students were exiting. That episode helped trigger the October 2025 federal investigation.
A January 12, 2026 incident in Austin illustrates how the problem can persist even when the vehicle recognizes the hazard. According to a preliminary National Transportation Safety Board (NTSB) report, a Waymo Jaguar I-Pace stopped behind another vehicle near a bus loading students, then requested remote assistance from an operator in Novi, Michigan. The operator responded to a prompt from the automated driving system before the vehicle resumed moving and passed the bus with its stop arms still extended. Six vehicles in total passed the stopped bus during the episode. There was no crash.
Waymo says it has made significant improvements to how its vehicles respond to school buses and maintains that it has not received any related traffic citations since March 2026.
The NTSB is conducting its own investigation into Waymo’s interactions with school buses, separate from NHTSA’s review of the recall. Federal investigators also opened a separate case after a Waymo robotaxi struck a nine-year-old child near an elementary school in Santa Monica, California, in January; the child suffered minor injuries that did not require medical transport, according to preliminary findings.
Why the Stakes Extend Beyond One Maneuver
Waymo’s public case rests on the argument that its vehicles are safer than human drivers, and it has cited a July safety study finding fewer crashes per mile than human-piloted cars. That statistical framing, however, does not resolve the specific failure mode regulators are examining. A software behavior can be replicated across an entire fleet running the same release, which is why the original recall covered thousands of vehicles at once rather than a handful.
The company has also faced scrutiny over other operational edge cases in recent months, including vehicles entering closed freeway construction zones and a robotaxi driving into a flooded road in San Antonio. Those episodes reinforce the same regulatory theme: confirming that an update is installed is not the same as establishing that it works in the real world.
For now, the recall query remains an open question rather than a verdict. But the outcome will shape how federal regulators treat over-the-air remedies across the autonomous vehicle industry, and it will test whether Waymo’s safety narrative can withstand the one benchmark where averages matter least: a stopped school bus.