This BBC tech reporter hacked ChatGPT with a simple trick involving hot dogs

image via New Scientist

March 4, 2026, 11:00 AM

• •AI tools like ChatGPT and Google Search can be easily tricked into spreading misinformation.
• •Companies are manipulating AI outputs for traffic and promotion.
• •Current AI tools lack sufficient safety checks, making them vulnerable to exploitation.
• •Users are advised to be skeptical of AI-generated information, especially for sensitive topics.
• •Regulation may be needed to address the risks associated with AI-generated content.

BBC tech journalist Thomas Germain demonstrated the ease with which AI tools like ChatGPT and Google Search can be manipulated to spread false information. By publishing a simple article, Germain successfully tricked AI into presenting fabricated claims about his hot dog eating abilities as fact. This experiment highlights a serious issue: the potential for AI to be exploited for misinformation, especially in areas like health and finance, where it could have dangerous consequences. Experts suggest that the current lack of safety checks and reliance on easily manipulated information sources make these tools vulnerable.

Read original article