In a statement, Waymo said an investigation into the incident is ongoing, and an initial review found the system correctly identified the dog but “was not able to avoid contact.”

“We send our sincere condolences to the dog’s owner,” a Waymo spokesperson told Fortune, acknowledging that the company did not know who the actual owner is.

“The trust and safety of the communities we are in is the most important thing to us, and we’re continuing to look into this on our end,” the spokesperson said. The death of someone’s beloved Fido comes as Waymo, a subsidiary of internet giant Alphabet, and its competitor, GM-owned Cruise, seek to operate self-driving taxis at all hours in the city. 

The incident happened around midday on May 21, while a Waymo Jaguar I-Pace was driving in autonomous mode around San Francisco’s Produce Market neighborhood at a speed of roughly 23 miles per hour. The dog ran off leash, out of a homeless encampment on Toland St., and made its way toward the side of the vehicle, the spokesperson said. The safety driver, who was seated behind the steering wheel but was not operating the car, reported being unable to see the dog as it ran out from behind a parked vehicle, and a collision followed. A report from California’s Department of Motor Vehicles noted that the car sustained minor damage while the “small dog,” listed in the section for property damage, “did not survive.”

Waymo has done further analysis to measure how a human driver would have performed under the same circumstances using a model known as NIEON. It was made to help determine readiness of autonomous driving systems and compare them to human drivers, essentially serving as a stand-in for the highest performing human driver that’s always attentive and doesn’t get fatigued. 

In a benchmark study last year, the Alphabet subsidiary found that its tech outperformed its NIEON driver in simulated fatal crash scenarios by avoiding more collisions and mitigating serious injury risk.

Since the fatal collision with the dog, Waymo has reviewed the incident from different perspectives, one of which involved reconstructing it in a simulation to compare the system’s performance to that NIEON, and determined the crash wasn’t possible to avoid. 

In a 2022 post about its safety assessments, Waymo shared its hope that sharing its methodologies would encourage discussions about safety metrics across the autonomous driving industry. 

Still, San Francisco transportation officials have called for more transparency about the operational data, after the city witnessed a spate of incidents, including a Waymo vehicle holding up traffic after remaining stopped in an intersection.  

Currently, GM’s Cruise is allowed to charge fares for driverless taxi rides between 10 p.m. and 6 a.m. in some parts of the city while Waymo can only charge fares if a safety driver is in the car. But both Waymo and Cruise are eying expansions of their autonomous vehicles.

In Texas, Cruise is expected to start testing its 6-passenger Origin shuttle, which has no steering wheel, brakes or other human controls on public roads in the state’s capital city. And in California, the public utilities commission will be voting near the end of the month on draft resolutions that, if approved, would grant the companies the ability to provide 24-hour robotaxi services throughout all of San Francisco.  

But officials in San Francisco expressed doubts about the safety of such a plan, requesting a more incremental approach in a January protest letter. 

Since then, incidents involving Cruise and Waymo vehicles have been “rapidly increasing,” a trend that follows a hike in driverless operations from Waymo early this year, city officials say

In a letter to the CPUC, transportation officials with the city described the resolutions as “backwards” and going against the commission’s duty to protect passengers and the general public. The group instead recommended more reporting requirements for the companies and adopting more minimum performance standards before all-hours robotaxi services are approved. 

Adding their hope that automated driving will be safer than human driving at some point, for now the officials say the technology is still “under development and has not reached this goal.”