It Was Frighteningly Easy to Make AI Believe a Hot Dog Lie

In the time it takes to boil a pot of pasta, a BBC tech reporter says he managed to rig what leading AI chatbots say about him. In a piece for the BBC, senior technology journalist Thomas Germain describes how he posted a fake article on his personal site declaring himself the top hot-dog-eating tech journalist, complete with invented contests (the 2026 South Dakota International Hot Dog Championship) and competitors (some also made up, others fellow journos who gave him the OK).

It took 20 minutes to write, and within a day, Germain says Google's Gemini, its AI Overviews, and ChatGPT were confidently repeating the made-up claim; only Anthropic's Claude resisted the lure of the frankfurter fraud. Germain and SEO experts tell the BBC this isn't just a silly stunt. The same simple tactic—posting persuasive content on your personal or company website, or getting a press release online via a paid-for distribution service—can be, and already is, used to skew AI answers on weightier topics like medical products, retirement investments, and local services.

Germain writes that people may be "more likely to fall for it" than they would have previously. Clicking on a search result gives you a chance to get a sense of whether the resulting website is clearly biased or just smells off. Now, the answer seems to come directly from leading tech companies, which can impart an air of authoritativeness. Read the full article for more.