One issue is that news sites article content is written to maximise ad impressions. Ads inject between paragraphs typically so you need lots of paragraphs. You start with perhaps two paragraphs of actual information, then break it down into 4 or more and add more information about related topics. Add some opinions, maybe weave in links to related articles on your site, and you end up with 10 or so paragraphs insterspersed with ads and pictures. You can remove the ads and pictures but you can't remove the bullshit. We need a good AI tldr machine.