May 20, 2026
Hot dog champ? AI says yes
Google's AI is being manipulated. The search giant is quietly fighting back
One fake blog post fooled AI, and commenters are calling the whole internet a scam
TLDR: A BBC test showed that a single fake webpage could trick major AI tools and Google’s answer box into repeating false claims, and Google is now scrambling to tighten the rules. Commenters were brutal, calling AI unreliable, internet-trained nonsense and warning that people are trusting shaky one-source answers too easily.
Google may say it merely “clarified” its rules, but the crowd is absolutely not buying the calm, corporate vibe. The real bombshell from the BBC investigation is that one silly blog post was apparently enough to make major chatbots repeat total nonsense — including the claim that a journalist was a world-champion hot-dog eater. Funny? Yes. Reassuring? Not even a little. In the comments, people sounded equal parts amused, furious, and exhausted, with one person groaning that the “llm-ification of every piece of text on the internet is driving me crazy,” basically summing up the mood of everyone who misses the old days of clicking actual links and deciding for themselves.
The hottest take was also the bluntest: “AI is such garbage. You can't use it for anything.” That scorched-earth verdict got backup from commenters comparing internet-trained AI to writing a school paper with the National Enquirer as your source. Ouch. Others were less ready to throw the whole thing in the trash, but demanded giant warning labels when an answer comes from one weird, obscure webpage. And honestly, the biggest gasp in the thread was over just how easy the trick sounded. One post? That’s it? For many readers, that turned this from a nerdy glitch into a full-blown trust crisis: if chatbots can be nudged on health, money, and search results this easily, then the joke isn’t the hot dogs — it’s the idea that the answer box knows best.
Key Points
- •A BBC investigation found that AI systems including ChatGPT, Gemini and Google AI Overviews could be manipulated by simple online content into producing misinformation.
- •The reporter says a single post on a personal website was enough to get some major AI systems to repeat a false claim.
- •The article says the vulnerability is especially exposed when AI tools search the web for answers and rely on limited or weak sources.
- •Google says its policy update is a clarification of existing anti-spam protections, though the article reports that companies appear to be increasing efforts behind the scenes.
- •The article says manipulated AI answers can affect decisions in areas such as health, finance, voting and consumer purchases.