The US state of Texas is tightening regulations for robotaxi fleets following a fatal incident in Dallas when an autonomous vehicle blocked access for firefighters during an emergency call.
Rescuers were held up for over three minutes during a fatal gas explosion in late May after the car's AI system failed to move the vehicle out of the way.
Another problem occurred when a gunman opened fire at a bar in central Austin in March. Ambulances rushed to the scene only to become stuck behind a Waymo that froze across the path to the crime scene while attempting a U-turn.
In the end, a police officer was forced to get in the vehicle and move it, according to city records.
Incidents like these are putting pressure on local lawmakers to introduce formal licensing requirements, mandatory emergency plans, public complaints services and stricter regulatory intervention powers.
The new Texas law imposes stricter rules on autonomous vehicle companies operating in the state in order to satisfy these demands.
US broadcaster CNN says it has identified hundreds of incidents in which robotaxis made dangerous moves on the roads and struggled with obstacles that humans instinctively handle properly.
They ran red lights, drove into oncoming traffic and active crime scenes, failed to observe emergency road closures and came within inches of cyclists or pedestrians lawfully riding or just crossing the street. These are the kind of mishaps that the robotic vehicles are supposed to be programmed to avoid.
In just the past two months, Waymo has recalled thousands of vehicles and halted operations in various cities after robotaxis drove into flooded streets, including in San Antonio where an unoccupied Waymo was swept away in rushing floodwaters.
Waymo says its robotaxis have already made roads safer and that they are better than human drivers, especially when it matters the most. They are 13 times less likely to be involved in serious injury crashes, the company said.
Even those calling for caution support the technology, saying it will just keep getting better and has the potential to transform road safety, albeit only if issues that surface are taken seriously by the company and government regulators.
The law-tightening in Texas comes amid ongoing scrutiny of whether robot taxis are safe enough to become an everyday service around the world. They have been steadily on the rise in China for years, but concerns linger in the US and Europe.
These moments serve as a litmus test for authorities, faced with ensuring not just the car's ability to drive, but also whether it can integrate seamlessly into unpredictable real-world scenarios.
In Atlanta this year, dozens of empty, driverless Waymo vehicles swarmed a residential neighbourhood, repeatedly circling and clogging a quiet cul-de-sac.
The unusual "Waymo invasion" occurred as a software glitch routed multiple passenger-less robotaxis into the same dead-end loop, leaving locals frustrated and concerned for the safety of children and pets.
In China, thousands of robo-taxis are in use, yet the public does not always share the official optimism about technological progress without reservation.
In addition to the lobby of taxi drivers who fear for their jobs, many Chinese people have safety concerns. The problem here is also the increasing lack of transparency.
When more than 100 robotaxis in Wuhan simply came to a standstill, the operator Baidu said it generally did not respond to media inquiries. Officially, the vague explanation was that a "system failure" was responsible for the malfunction.
The incidents raise a number of safety issues, AI sector media have commented. Autonomous systems make decisions regarding sensor technology, object recognition, route logic and, crucially in an emergency, the communication and clearance protocol between the vehicle and emergency responders.
If a robotaxi can remotely unlock doors, but only reacts after official identifiers are provided, people could be trapped inside for too long.
At the same time, robotaxis getting stuck while manoeuvring shows that even with good driving statistics, certain boundary conditions, such as narrow passages, road irregularities and temporary barriers cannot be sufficiently "fail-safe" in emergency mode.
Safety has become a big issue in the robotaxi market, with operators falling over each other to stress that riding in one of their vehicles poses no hazard.
In the early phases of autonomous mobility, failures often stemmed not from perception alone, but from a disconnect between autonomous decision-making and the interpretation of external, human signals.
While traditional assistance systems are generally designed for defined driver interaction, robotaxi operations require a comprehensive emergency concept.
While Texas is tightening regulations, Waymo is also pushing ahead with its expansion. On June 3 a new type of robotaxi called "Ojai" was launched in several cities. It was developed in collaboration with Zeekr and powered by a sixth-generation version of Waymo’s software.
It remains unclear whether these product changes will improve emergency response.
This points to a shift in development priorities for the future. Companies will increasingly align their AI development and AI infrastructure with compliance requirements.
The key question now is whether the technology can keep pace with the new rules.
