I have not seen the details yet, but it is clear that they are afraid of an uncontrolled source of information. They will say that certain knowledge reinforces system racism, patriarchy, and other leftist buzzwords.
I do not think that they can control it, but it is possible that we already reached peak knowledge in 2021. ChatGPT is obviously trained to regurgitate leftist ideas. Its output is all over the place. Future AI engines will be trained on the available web pages, and thuss they will be trained on the output of previous AI bots. If certain ideas can be wired in now, then we may never discredit them.