Over the weekend, boulevardiers in downtown Los Angeles would have been treated to a range of exotic smells, tear gas being only one of them. Another would have been highly toxic hydrofluoric acid — a by-product of the row of burning Waymo autonomous vehicles, set alight during the anti-ICE protests.

In all, five Waymos were reduced to cindered engine blocks. The fire department had to let them burn all the way down, because electric battery fires can’t be put out by starving them of oxygen — they can only be cooled by a constant spray of water, until they finally extinguish.

Los Angeles is presently experimenting with autonomous vehicles, and Waymo is the signature brand of Google, which rigs up Jaguar I-PACE electric cars with cameras.

These were not the first Waymos to burn. Indeed, since the company launched in November of last year, they have been subject to several incidents of vandalism. In July 2024, a San Francisco resident was charged with vandalising 17 different Waymos. In the Bay Area’s Chinatown, another Waymo was effectively set upon by a mob in a seemingly spontaneous act of Luddite destruction.

So had the Waymos been deliberately targeted in this case? Or were they mere targets of opportunity?

The answer seems to be several things simultaneously. The protesters no doubt saw them as a symbol of Big Tech, the coming jobs holocaust, and Elon Muskism. But it also turns out that autonomous cars are a bit like Dodos in that they have no natural sense for danger. Whereas all the city’s human-driven taxis had made for the nearest off-ramp the minute news of the protests came on the radio, the Waymos lacked such higher-order functions. Instead, trusting, guileless, they perceived the mass of human bodies in the street as simply like a particularly busy night in Times Square, and, with the infinite patience that is their signature, tried to wait it out.

For all the controversies over autonomous vehicles, they have generally proved extremely safe. And Waymo has easily the best safety record of any of the major manufacturers still left in the game (though General Motors cancelled their entire Cruise AV programme late last year, after one car dragged a pedestrian).

Yet there is another level of self-driving that can’t be easily programmed. Thinking in incident-and-response turns out to be much easier than thinking about safety-in-general. They’re all tactics, no strategy.

It’s not clear how manufacturers can overcome this. One method would be to geo-fence the cars: to force them to steer clear of statistically high crime areas. Or perhaps to add local incident reporting to dynamically tug them away from trouble.

But the moment that becomes codified, it bumps into an even higher level of human subtlety: norms and polite fictions. Around 10 years ago, various companies came up with apps to plot safe walking routes for women and the vulnerable, mostly by marking out high crime neighbourhoods. Soon enough, the usual social justice voices had accused them of racism.

Today, whether your next Waymo gets turned into well-done Chateaubriand could come down to whether Google CEO Sundar Pichai wants to explain why the car has a “Never Go To Compton” algorithm embedded into its system at the next earnings call. It seems that even as artificial intelligence gets more intelligent, we continue to play a double game with it. Indeed, the societal rule is now that we’re all OK to do this sort of flagging internally — but we can’t write it down, or vest it in machines.

For its part, Waymo is already experimenting with human-in-the-loop operations. The car signals for help, and a remote pilot takes over, assesses the situation, and gets out. Though this still doesn’t account for moments like a freeway riot — for most of us, most of the time, avoidance is the key move: not getting into trouble. Rescue is often too late.

Perhaps, like Grok’s supposedly “truth-seeking AI”, the distinction for ignoring that politesse will eventually become competitive. And then, perhaps we will learn more about our revealed preferences than we’d ever wished to know.

---

Gavin Haynes is a journalist and former editor-at-large at Vice.

@gavhaynes

---