http://damaga37e3smvdiz473qnzyfnul275erstpaihy2apvwq3bspjxyvnad.onion/git/anonim/telegram-tampering/src/branch/main/article-en.md
First, it checks, then, if there are certain triggers / quantities and qualities of complaints about the content, it transfers the content for processing by a human, if the AI itself cannot determine whether the content is legal or not. The AI copes well with spam, obviously spam and fake accounts are removed FROM SEARCH almost immediately after a regular report made through the button.