What are their other options? Imagine they could, with great accuracy, predict whether somebody is going to leave a good comment or a bad one. Do they expose that to the commenter, saying they've been blocked from commenting because the algorithm thinks they're a shitty person? Imagine the optics of that...
The other option is the shadowban, where toxic commenters are hidden without notice. It sounds like a good idea, but just about led to a user revolt when Reddit tried it.
Simply turning off comment sections that have the potential to become toxic may not be the most technologically interesting solution, but it's the safest PR move.