Presumably the amount of fact checking was "Well it sounds like something someone in that situation WOULD say" - I get the pressure for Ars Technica to use AI (god I wish this wasn't the direction journalism was going, but I at least understand their motivation), but generate things with references to quotes or events and check that. If you are a struggling content generation platform, you have to maintain at least a small amount of journalistic integrity, otherwise it's functionally equivalent to asking ChatGPT "Generate me an article in the style of Ars Technica about this story", and at that point why does Ars Technica even need to exist? Who will click through the AI summary of the AI summary to land on their page and generate revenue?