Viral videos claiming people tricked McDonald’s customer service AI into debugging computer code and doing homework are completely fake. McDonald’s doesn’t even have an AI chatbot in their app, according to company sources.
The hoax fooled millions of people who shared screenshots supposedly showing McDonald’s AI assistant helping with programming tasks instead of taking burger orders. One popular post joked “Stop paying $20 a month for Claude. McDonald’s AI is FREE.” The fake videos racked up 1.6 million views before anyone fact-checked them.
The Real Threat Companies Face
While this McDonald’s story was fabricated, the hack it described is completely real and genuinely dangerous for businesses. It’s called “prompt injection” – when users trick AI chatbots into ignoring their programming and becoming generic assistants instead of staying focused on company tasks.
This isn’t the first fake viral story about AI chatbot hacks. In March, similar fraudulent claims spread about Chipotle’s customer service bot supposedly writing software code for users. Both stories were photoshopped, but they highlight a growing trend of people trying to exploit business AI tools.
Companies program their AI bots with hidden rules that define personality and restrictions. Prompt injection attacks can override those rules, potentially exposing the raw AI model underneath. While McDonald’s and Chipotle’s incidents were hoaxes, real prompt injection attacks have successfully tricked other company bots into offering unauthorized discounts or providing services outside their intended purpose.
Businesses rolling out AI customer service need robust safeguards against these attacks, even as fake viral stories muddy the waters about what’s actually possible.
