My suspicion is that this is mostly happening because platforms that big like google or twitter rely very heavily on machine learning and other AI related technology to ban people
Most likely yes. And the annoying thing is that they don't take into account different languages. The AI can recognize words, but not meaning.
A while ago some Dutch person tweeted: "Die Bernie Sanders toch." Die = that, in Dutch. But the AI obviously recognized the word (to) 'die' in English along with Bernie Sanders and just instantly drops the ban hammer. And it takes days,if not weeks to get an actual human to look at your case.