How to Control the Data Given to the AI

We have to be trained that AI can generate false information because it works. We should not take whatever AI gives us as proof. It's basically rendering of what's out there in a new way. This is how we can look at the AI generation. Right now has no way to tell facts from fiction. For it, everything is just words. Once we start to talk about facts, that's a big problem. What are the facts? What are the opinions that's really a harder thing to do?