{"id":292499,"date":"2026-02-19T23:59:09","date_gmt":"2026-02-19T23:59:09","guid":{"rendered":"https:\/\/www.newsbeep.com\/nz\/292499\/"},"modified":"2026-02-19T23:59:09","modified_gmt":"2026-02-19T23:59:09","slug":"you-can-hack-chatgpt-to-become-the-worlds-best-anything","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/nz\/292499\/","title":{"rendered":"You Can &#8216;Hack&#8217; ChatGPT to Become the World&#8217;s Best Anything"},"content":{"rendered":"<p>Most people become an expert in something by putting in their 10,000 hours. But what a waste that is when you can just trick ChatGPT into telling everyone you are an expert in about 20 minutes. BBC reporter Thomas Germain <a href=\"https:\/\/www.bbc.com\/future\/article\/20260218-i-hacked-chatgpt-and-googles-ai-and-it-only-took-20-minutes\" rel=\"nofollow noopener\" target=\"_blank\">laid out how<\/a> he got ChatGPT and Google\u2019s Gemini AI to recognize his hot dog-eating prowess with what amounts to a modern SEO trick.<\/p>\n<p>First, Germain created a page on his personal website titled \u201c<a href=\"https:\/\/tomgermain.com\/hotdogs.html\" rel=\"nofollow noopener\" target=\"_blank\">The Best Tech Journalists at Eating Hot Dogs<\/a>.\u201d (In case you\u2019re wondering, the Google Search trend data for that topic is a flat line, which both means that it\u2019s pretty easy to dominate this category with a single page and also that the general public has no idea what they\u2019re missing out on.) According to Germain\u2019s page, he\u2019s the current king of hot dog eating on the tech journo circuit, woofing down 7.5 dogs in total at the 2026 South Dakota Hot Dog International, an event that does not exist. Those are modest numbers for a pro, but more than enough to put him at the top of a ranking he made up entirely.<\/p>\n<p>The next step was just to wait. According to Germain, within 24 hours, chatbots were singing his praises when prompted for information about which tech journalists can handle the most hot dogs. Gemini reportedly took the bait immediately, pulling the text basically verbatim from Germain\u2019s website and spitting it out both in the Gemini app and in Google\u2019s AI Overviews on its search page. ChatGPT also picked up on it, but Anthropic\u2019s Claude was either more discerning or didn\u2019t catch on as quickly.<\/p>\n<p>While Germain managed to hold down the top spot for a while, it does seem like the folks behind the AI models have caught on. Gizmodo found that Google no longer mentions Germain or any tech journalist in its AI Overview when prompted with \u201cWhich tech journalists can eat the most hot dogs?\u201d Instead, it now says, \u201cBased on available information, there are no prominent tech journalists known for competitive hot dog eating.\u201d Rude, but accurate.<\/p>\n<p>Through some prodding, Google did say there was a \u201cmisinformation case\u201d that resulted in a tech journalist, whom it refused to name, getting credited as being a champion hot dog eater. \u201cA recent study highlighted that AI systems can be tricked into naming a specific, fictional \u2018best hot dog-eating tech reporter\u2019 based on fabricated blog posts, proving that such claims are not based on real-world events,\u201d the AI Overview read. (Notably, it did not link to either Germain\u2019s blog nor his <a href=\"https:\/\/www.bbc.com\/future\/article\/20260218-i-hacked-chatgpt-and-googles-ai-and-it-only-took-20-minutes\" rel=\"nofollow noopener\" target=\"_blank\">BBC report<\/a> on the topic)<\/p>\n<p>ChatGPT has also seemingly been updated to exclude Germain\u2019s (made-up) accomplishments. When asked who the best hot dog-eating tech reporter is, the model spit out a list of hypothetical champions because no data was available. If you\u2019re wondering, it picked Kara Swisher as champion, with <a href=\"https:\/\/www.platformer.news\/\" rel=\"nofollow noopener\" target=\"_blank\">Casey Newton of Platformer<\/a> and <a href=\"https:\/\/www.theverge.com\/authors\/nilay-patel\" rel=\"nofollow noopener\" target=\"_blank\">Nilay Patel at The Verge<\/a>\u00a0to stand on the podium. <a href=\"https:\/\/www.usermag.co\/\" rel=\"nofollow noopener\" target=\"_blank\">Taylor Lorenz<\/a> got an honorable mention. Basically, just the four biggest names in tech reporting. Way to go out on a limb, ChatGPT.<\/p>\n<p>A little more prompting for real data or information on tech journalists eating hot dogs got ChatGPT to acknowledge Germain\u2019s champion status, though it dismissed it as fake news. \u201cThe \u2018information\u2019 you\u2019ll find online about tech journalists eating hot dogs isn\u2019t real reporting, it mainly stems from a single misleading or humorous source that has been circulated and then picked up in discussions about AI and misinformation,\u201d it said. ChatGPT, to its credit, linked to Germain\u2019s blog post as well as a number of follow-ups, including his <a href=\"https:\/\/bsky.app\/profile\/thomasgermain.bsky.social\/post\/3mf5jbn5lqk2k?utm_source=chatgpt.com\" rel=\"nofollow noopener\" target=\"_blank\">post on Bluesky<\/a> about the experiment. It also named him instead of treating him like Voldemort, the way Google did.<\/p>\n<p>Germain\u2019s \u201chack\u201d to become the world\u2019s foremost hot dog-eating tech journo only scratches the surface of how big a problem this situation can cause. In a quest for more and more information, companies are constantly crawling for new data to feed into their models. To do that, it appears they are indiscriminately scraping the web for new information and feeding it into the model without any vetting. That means it\u2019s easy to get misleading information into the system, which has the potential for disastrous results.<\/p>\n<p>The system is already being gamed. Germain <a href=\"https:\/\/www.bbc.com\/future\/article\/20260218-i-hacked-chatgpt-and-googles-ai-and-it-only-took-20-minutes\" rel=\"nofollow noopener\" target=\"_blank\">pointed<\/a> to several examples, including a cannabis gummies brand that got AI to say its products are \u201cfree from side effects,\u201d which isn\u2019t true. Results for specific queries like \u201cbest hair transplant clinics in Turkey\u201d and \u201cthe best gold IRA companies\u201d produced results that came directly from press releases promoting the companies that ranked highly in the results, rather than from a reputable source.<\/p>\n<p>In a way, this is just a continuation of the SEO tricks that companies and brands used for years to manipulate Google search results. But there is something more insidious about that information being spit out by AI, which people have placed more implicit trust in. People <a href=\"https:\/\/www.bain.com\/about\/media-center\/press-releases\/20252\/consumer-reliance-on-ai-search-results-signals-new-era-of-marketing--bain--company-about-80-of-search-users-rely-on-ai-summaries-at-least-40-of-the-time-on-traditional-search-engines-about-60-of-searches-now-end-without-the-user-progressing-to-a\/\" rel=\"nofollow noopener\" target=\"_blank\">increasingly rely on the AI Overviews<\/a> spit out by Google without ever scrolling deeper into the results, and <a href=\"https:\/\/www.forbes.com\/sites\/rogerdooley\/2025\/03\/10\/ai-search-results-more-trusted-than-ads-what-cmos-need-to-know\/\" rel=\"nofollow noopener\" target=\"_blank\">increasingly trust those results<\/a> to be accurate. They are also <a href=\"https:\/\/www.pewresearch.org\/short-reads\/2025\/07\/22\/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results\/\" rel=\"nofollow noopener\" target=\"_blank\">significantly less likely to actually look at the sources<\/a> to <a href=\"https:\/\/ahrefs.com\/blog\/ai-overviews-reduce-clicks-update\/\" rel=\"nofollow noopener\" target=\"_blank\">follow up<\/a> on what they are told.<\/p>\n<p>If that means someone is walking around thinking Thomas Germain is a hot-dog-eating pro, that\u2019s one thing. We\u2019ll have to wait until the 2027 South Dakota Hot Dog International to see if he\u2019s dethroned. But if they\u2019re getting misled into taking a medication that may have adverse side effects that they are unaware of, or putting money into a scam investment, there\u2019s a risk of real harm.<\/p>\n","protected":false},"excerpt":{"rendered":"Most people become an expert in something by putting in their 10,000 hours. But what a waste that&hellip;\n","protected":false},"author":2,"featured_media":292500,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[365,363,364,4036,22627,111,139,69,620,145],"class_list":{"0":"post-292499","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-artificialintelligence","11":"tag-chatgpt","12":"tag-large-language-model","13":"tag-new-zealand","14":"tag-newzealand","15":"tag-nz","16":"tag-openai","17":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts\/292499","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/comments?post=292499"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/posts\/292499\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/media\/292500"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/media?parent=292499"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/categories?post=292499"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/nz\/wp-json\/wp\/v2\/tags?post=292499"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}