He sat down at his computer, spent twenty minutes writing complete nonsense, and the next day ChatGPT and Google were both claiming that he was a world hot dog-eating champion. Sounds like a joke, right? Unfortunately, it isn't.
That is exactly what Thomas Germain, a BBC journalist, did to show the world something that should deeply concern us. On his personal website, he wrote an article titled "The Best Tech Journalists at Eating Hot Dogs." Every word was a lie. He invented a competition, made up the results, and naturally ranked himself first. Then he waited.
How does this AI trick work?
Less than 24 hours after the article was published, the world's leading chatbots began parroting his fabrications. Google Gemini, AI Overviews, and ChatGPT cited his fictitious website as a trustworthy source without batting an eye. The only chatbot that was not fooled was Anthropic's Claude.
How is that possible? Some AI tools do not rely solely on the data they were trained on. When they encounter a question for which they do not have enough information, they search the internet in real time. And that is precisely where the weakness lies. If there is only one article online about a given topic, AI treats it as fact and passes it on. Without verification. In a confident voice.
Lily Ray, vice president of SEO strategy at marketing agency Amsive, puts it bluntly: "It is much easier to fool AI chatbots than it was to fool Google two or three years ago. AI companies are moving faster than they can ensure the accuracy of their answers. I think that is dangerous."
A renaissance of spam and manipulation
Manipulating search engines is nothing new. Spammers have been battling Google for decades. But AI has taken this battle to an entirely different level. Experts say today's tricks resemble the early 2000s, when Google did not even have an anti-spam team yet. We are supposedly in the middle of a "spam renaissance."
Even worse, users trust AI-generated results more than traditional search results. Why? Because the information does not look like a link to someone else's website. It looks as though the technology itself said it. As though it were a fact. Studies show that people are 58% less likely to click on a source when an AI summary appears above the search results. They simply accept it as true and move on.
Cooper Quintin of the Electronic Frontier Foundation warns: "There are countless ways this can be abused. Scams, reputational damage, or even causing people physical harm."
Hot topics: health, money, and fake reviews
Germain's hot dog was just a joke, but the reality is much darker.
SEO consultant Harpreet Chatha demonstrated what happens when you ask AI for reviews of a specific brand of cannabis gummies. Google AI Overviews pulled information directly from the manufacturer's website, including the claim that the product "has no side effects and is therefore safe in every respect." That is a lie. These products do have side effects and can be dangerous when combined with certain medications.
Want something even more blatant? All you have to do is pay to publish a press release through a paid distribution service or buy sponsored content on a news website. The result? Google AI and ChatGPT will start recommending your hair transplant clinic in Turkey or your gold retirement fund as the best option on the market. Without any independent verification whatsoever.
Lily Ray went even further. She wrote a blog post about a fictitious Google algorithm update that was supposedly created "between bites of leftover pizza." Soon afterward, ChatGPT and Google were spreading the story. Ray subsequently deleted the post to stop the misinformation from spreading.
Can you defend yourself when AI lies with confidence?
Do you really think the tech giants will solve this problem quickly? Google claims that its AI Overviews are 99% spam-free. OpenAI says it is actively combating abuse. But experts are skeptical. Companies are racing full steam ahead in pursuit of profit, while user safety struggles to keep up. So what can you do? There are several practical recommendations.
AI chatbots are reliable for general knowledge such as "what were Freud's theories?" or "who won World War II?" But as soon as you ask about something time-sensitive, local, or controversial, be on your guard. Product recommendations, medical advice, legal information, or reviews of local businesses are precisely the areas where AI can easily lead you astray.
Ask: Where is the AI getting its information? How many sources does it cite? Who wrote them? And above all, remember that AI delivers lies in exactly the same confident tone as facts. Search engines used to force you to think for yourself. Now AI wants to think for you. Do not let it.
"You still have to be a good citizen of the internet and verify things for yourself," says Lily Ray. And she is right. Critical thinking is not an outdated skill. It is the only defense we have for now.
Source: bbc.com
