Because the project is small right now (and makes no money), I've included a very limited list of sources for each category, however at scale we would have alot

Because the project is small right now (and makes no money), I've included a very limited list of sources for each category, however at scale we would have alot of options for getting past this. For example, one way in theory is to scrape all known publicly available news sites and create an article for every single detected story, but this would be very expensive. As someone with experience in the media, what steps would you take to find these under-reported stories? Maybe the steps can be replicated with AI to increase coverage?