But they are also trying to protect others from you. Just because you want to post videos of beheadings and things doesn't mean others want to accidentally bump into it?
I'd have found this a much stronger argument if the author hadn't dismissed words altogether. There's a huge gap between "I know where you live <address> and I'm going to set your house on fire while you sleep" and "I think there are only two genders", and I think the author only has the latter in mind. I think I'd disagree even then, but much less strongly.
The related rhyme, "Sticks and stones may break my bones, but words shall never hurt me" is similarly ridiculous. Words can totally hurt people. People have been driven to suicide, and to warfare, with words.
People invoke safety for the same reason a thousand years ago people invoked god. It's a religion, it's dogmatically considered good by default and it takes massive effort to argue against it.
Some manipulating jerk can invoke the name of either with a few simple words and anyone who wishes to call them on their moral bankruptcy needs to spend orders of magnitude more words doing so if they are to produce a rebuttal that cannot be trivially dismissed, bogged down or diverted.
And unless we as a society are prepared to punish people for employing these rhetorical tricks it will keep happening.
And FWIW I'm addressing the author's point in general, not in this specific case of a statement made by a company. I'm aware that when companies say safety they usually mean "their safety from the mob or the state or the ambulance chasing lawyers"
It has certainly negatively polarized people against any form of safety near AI.