All of the big tech social media began using AI for moderation by around 2020. It was always advertised as stopping spam and “inauthenticity” but was abused from the beginning as a no-appeal means of silencing those the staff disagreed with. There are undercover videos from years ago where the people who used to run Twitter talked about it and it also showed up in the source code they put on the web after Musk took over. They had undesirable topics that flagged people to be shadowbanned. The general idea was that if whatever you talked about might offend the Senior VP in the corporate lunchroom then you were eventually flagged and made invisible.

Now it’s less obvious because they use ghost banning where your tweets and replies seem to show up, and even get a handful of views from your existing followers, but you’re 100% invisible to the rest of the world. “We put a label on your account” appears in your messages a short while after the ghost ban begins at X. If you reply a lot, it seems guaranteed. Many people have no idea they’re invisible to the world, although when you never get a reply or a like, you’re not as inclined to keep using the platform, so the result ends up being the same as if they did find out.

I think they do it now because they’ve downsized and don’t have many real people working there. The other motivation may be they want to systematically keep the vocal little people from drowning out the commercial accounts who matter to them. Everyone else is there to consume content and view ads.

Earnings are way down from the peak years because the corporate social media enshittifiction drove the majority of the user base to video stream chats like Rumble, Discord, Twitch and youtube chats. Youtube doesn’t seem nearly as inclined to screw with that given the competition. It’s one of the small number of cash cows that pay for everyone else working on neat R&D projects that get cancelled without ever earning a dime.