"If he's done his own research which he has, how is a AI model going to fact check that?" - Where is this question coming from?
I assume with "he's done" you refer to Rupert Lowe? If so, what do you mean with "doing his own research"? Is doing his own research, to fabric his personal reality to please his article about it? Or do you also talk about researching data, backed from criminal cases, police reports and so on? Since in the second case, it would be important, that the report would state based on which dataset it was written and that those reports are verifiable. When the data are in an analog library be it. Then some people can still go there and look up the same data, as the author has seen.
Since a truthful report about whatever thing on earth can be checked by others. This is how trust is built. When many can check the data, verify the logic of the claims and independently come to the same findings.
How do you build your trust on literature?