Contrary to the wording of the report, the fake news detector isn't a neural net or an AI. The term is a bit overused, so it's not really worth it to be up in arms about the terminology. The tool is really just a weighted measure of a couple of online services that track website metadata. It would be interesting to use this to train a model with in order to flag potential fake news in the future if they decide to continue with the concept.

As it stands now though, it's not special in it's construction, but it's nice to see something on the more practical end emerge from a Hackathon.