They’ve always been strict at blocking potentially offensive text. Ever kept trying to compose a helpful, constructive YouTube comment reply that doesn’t disappear when you refresh? The AI there is really draconian. Once, someone wanted details of a rather wholesome story I had shared in a top level comment. Replying with the follow-up always failed despite no obvious trigger words. I resorted to editing the parent comment but then struggled to get a “see edit” reply through. At least 10 attempts at expressing “look at the root comment for updates” failed, eventually I managed to get a reply through with nothing but “🔝”. It got removed after I tried to edit it to a clearer expression so I just posted another “🔝” reply and hoped for the best.