It's kind of silly - it's not that the robot was bad but that if you keep it on 24/7 it will eventually say something bad. An infinite monkey typewriter will contain transphobic text.
15 years and people are still confused about this one?
They can't own accounts and they can't decide what content is shown any more than the rain can decide if my dog gets wet.
So the notion of people being absolved of responsibility for the procsses they set in motion is absurd.
One way or another, the concept of an out of control AI is not sustainable.
The era of "my chat bot does whatever it wants and I'm not liable for it" is likely going to end in the developed world over the next decade or two. You run it, you're responsible for it.
There was a Twitter account called "Fuck every word" which went through every word in the dictionary in order, writing a twit which consisted of "Fuck X" where "X" is the word.
It got promptly banned as soon as it reached certain word starting with N.
I imagine a sufficiently advanced AI or even human could create a convoluted word or phrase and breadcrumb enough to walk it back to whatever would have been censored, a la "for want of a nail" style.
If every strike on youtube was a permanent ban there would be many more creators on alternate platforms like rumble. They're not there to teach you, just to incentivize you to follow their rules or go somewhere else. Thus the complex system of strikes, demonetization, temporary live-streaming bans, etc.