If they have a machine which is good at identifying suspicious information, on a huge scale, they may also be able to control which information is suspicious, to a limited extent. It could easily be part of an effective apparatus of repression.
There's nothing in the article about a machine which identifies suspicious information, or information control, or an apparatus of repression. Perhaps you're referring to some other thread?
There is however, an open-source AI, and a github link to it. If you're familiar with Python you can literally read the code yourself and look for anything nefarious.
