Deepfakes, disinformation, controversial forums, banned books are all a threat to "democracy" and AI is making it worse, especially with AI agents (bots) that pretend to be users. X.com's Community Notes is good but it's powered by humans by design, and there's a limited number of humans of course, so I think it will fail against an unlimited number of AGI bots. Centralized social media will have to fight this by enforcing things like C2PA (in countries that will require it by law) and banning you if you don't comply, while p2p social media like plebbit can't ban you, so information will flow freely regardless of the outcome; it's effective accelerationism. I think most subplebbit owners will want real users, so they might end up using biometrics as anti-spam challenges, maybe something like worldcoin but hopefully free and open source.