Waymo’s robotaxis stalled during a San Francisco blackout, exposing critical gaps in autonomous vehicle emergency preparedness — and forcing regulators to confront whether robotaxis can truly handle disasters.
The December 20 power outage in San Francisco, triggered by a fire at a PG&E substation, plunged roughly one-third of the city into darkness. Videos circulating online showed Waymo robotaxis immobilized at traffic lights with hazard lights flashing — unable to proceed without human intervention despite being designed to navigate autonomously.
Waymo, operating over 2,500 vehicles across five U.S. cities including San Francisco, suspended operations for a day before resuming service. In a statement, the company attributed delays not to software failure but to “a concentrated spike” in confirmation requests from its fleet response agents — humans remotely monitoring vehicles — who needed time to verify safe passage through darkened intersections.
“The outage created a concentrated spike in confirmation requests that led to response delays contributing to congestion on already-overwhelmed streets,” Waymo said. This wasn’t an unexpected system crash — it was a predictable bottleneck under stress.
Autonomous vehicle operators rely on remote teleoperation teams to manage edge cases. Waymo’s fleet response agents, trained to interpret ambiguous scenarios, became overwhelmed when thousands of vehicles simultaneously requested verification while navigating unfamiliar conditions. The incident revealed a core flaw: autonomous systems are optimized for routine driving, not crisis management.
“If you get a response to a blackout wrong, regulators are derelict if they do not respond to that by requiring some sort of proof that the earthquake scenario will be handled properly,” said Philip Koopman, professor of computer engineering at Carnegie Mellon University and expert in autonomous technology. His words underscored what many experts now see as inevitable: robotaxis cannot operate safely without redundant, fail-safe protocols during emergencies.
Missy Cummings, director of George Mason University’s Autonomy and Robotics Center and former adviser to the U.S. National Highway Traffic Safety Administration, echoed this sentiment. “The whole point of having remote operations is for humans to be there when the system is not responsive in the way it should be,” she said. “The federal government needs to regulate remote operations… to ensure backup systems exist when catastrophic failure occurs.”
California’s Department of Motor Vehicles and Public Utilities Commission confirmed they are investigating the incident and formulating new regulations. “We’re talking to Waymo and other autonomous vehicle makers about actions related to emergency response,” a DMV official stated. “We’re also crafting rules to ensure remote drivers meet high standards for safety, accountability, and responsiveness.”
This isn’t the first time robotaxi deployments have faced scrutiny. After Cruise’s fatal accident in 2023, regulators revoked its permit — leading to its eventual shutdown. But Tesla’s rapid expansion in Austin and Waymo’s steady growth since its 2009 Google origins show the industry remains aggressive. Yet public trust continues to erode unless failures like these are met with clear, enforceable safeguards.
Waymo claims it’s refining its confirmation processes to better handle large-scale disruptions. “We’re implementing fleet-wide updates that provide vehicles with specific power outage context, allowing them to navigate more decisively,” the company said. But until these updates are validated under real-world stress tests — not just simulated ones — the public remains skeptical.
Experts argue that beyond software patches, robotaxi operators must face additional permitting requirements once fleets surpass certain thresholds. “If this had been an earthquake, it would have been a problem,” Koopman warned. “This is just a shot across the bow.”
For users, the takeaway is simple: autonomy doesn’t mean invincibility. Robotaxis may never fully replace human-driven cars — especially in unpredictable environments. For developers, the lesson is clearer: emergency protocols aren’t optional add-ons — they’re foundational infrastructure. Without them, even the most advanced AI becomes a liability during crises.
As Waymo resumes operations and competitors like Tesla and Zoox accelerate deployment, the pressure mounts on regulators to act. The next phase won’t be about speed or scale — it will be about resilience. And if history is any guide, those who fail to prepare now will pay dearly later.
Stay ahead of the curve — onlytrustedinfo.com delivers the fastest, most authoritative analysis of breaking tech news. Subscribe to our newsletter for daily deep dives into what matters most for users and developers alike.
