There is also always a chance that it's simply going to be wrong, ML cannot differentiate what is the truth or not. We see it happen with easy mistakes that people wouldn't make and it's going to be even worse when they get used for something more nuanced or complex.

Of course it’ll be trash. Quality isn’t the goal, just bulk with the aim of getting maybe fewer views per article but pumping out so so many that it’s more views, or rather ad impressions, overall with much lower cost.

Problem is it’s shortsighted. Once those sources quickly get a reputation for trash quality folk will learn not to bother clicking through to those sources.