> And Fakespot: they present themselves as a company that focuses on detecting AI-generated content from human-generated content. It sounds like they've set out

> And Fakespot: they present themselves as a company that focuses on detecting AI-generated content from human-generated content. It sounds like they've set out to play Whac-A-Mole against the all the biggest AI companies in the world. I use FakeSpot a lot, and it's less about finding AI written reviews and more about finding any fake reviews, human written or not. They don't need to try and tell if reviews are actually written by a human, just if they are authentic. A reviewer with only 1 review on their whole account will be flagged as suspicious. A reviewer with only reviews for the same company will be marked as definitely fake. It also does things like look for repeated sentiments across all reviews for a product. Sometimes it gets false positives, like a review for a chair might legitimately have 30% of users saying the phrase "super/very/really comfy" at some point in their review. But because it's repeated so much it's flagged as indicating fake reviews. There are lots of companies working on making convincing language models. ChatGPT is pretty much already at that level for something simple like Amazon reviews. But there aren't any large AI companies working on AI that can fake looking like an authentic group of reviewers. Those are all more shady businesses without billion dollar budgets. Sure the problem of writing a single convincing review is now solved for those shady businesses. But the really sophisticated ones were already paying humans to write the reviews before good LLM's came along. There's also the issue of sellers themselves including a card or followup email promising a small gift card in exchange for a 5 star review.