Yes. [1] That incident got considerable publicity in the San Francisco media. But not because of the Waymo.[2][3]

Someone was driving a Tesla on I-280 into SF. They'd previously been involved in a hit-and-run accident on the freeway. They exited I-280 at the 6th St. off ramp, which is a long straightaway. They entered surface streets at 98 MPH in a 25 MPH zone, ran through a red light, and reached the next intersection, where traffic was stopped at a red light. The Tesla Model Y plowed into a lane of stopped cars, killing one person and one dog, injuring seven others, and demolishing at least six vehicles. One of the vehicles waiting was a Waymo, which had no one on board at the time.

The driver of the Tesla claims their brakes failed. "Police on Monday booked Zheng on one count of felony vehicular manslaughter, reckless driving causing injury, felony vandalism and speeding."[2]

[1] https://www.nbcbayarea.com/investigations/waymo-multi-car-wr...

[2] https://www.sfchronicle.com/sf/article/crash-tesla-waymo-inj...

[3] https://www.youtube.com/watch?v=ULalTHBQ3rI&

The question should be less who was at fault and more would a human driver have reacted better in that situation and avoided the fatality. I'm not sure why you think that whether the fatality occurred inside or outside of the car changes the calculus, but in that case only one of the two documented Tesla FSD-related fatalities killed the driver. Judging by the incident statistics of Tesla's Autopilot going back over half a decade, I'm pretty sure it's significantly safer than the average human driver and continues to improve, and the point of comparison in the original post was with human driving rather than Waymo. I have no doubt that Waymo, with its constrained operating areas and parameters, is safer in aggregate than Tesla's general-purpose FSD system.

Only one of the two, and it's not nearly enough data to draw a conclusion one way or another in any case.

FSD is not Autopilot despite the names being conflated today, but even if you want to count all 28, it's not enough to compare raw numbers of fatal incidents without considering the difference in scale. That's not to justify taking your eyes off the road when enabling FSD on a Tesla, but the OP did not suggest that either anyway.

Hell, they'll even hold a press conference touting (selective data from) telemetry to say "The vehicle had been warning him to pay attention prior to the accident!"

And then four months later when the actual accident investigation comes out, you'll find out that yes, it had. Once. Eighteen minutes prior to the accident.

And then to add insult to that, for a very long time, Tesla would fight you for access to your own telemetry data if, for example, it was needed in a lawsuit against someone else for an accident.

There are news articles of people doing that on the 5 freeway