6 nominees for this year's inaugural AI Darwin Awards
These dubious honors will spotlight some of the dumbest things humans have done with AI this year
Moor Studio
The Darwin Awards, by their very definition, are analog. The tongue-in-cheek annual event centers around deaths that involve immensely poor judgment, removing that person from the larger gene pool. But just as artificial intelligence is seeming more humanistic, it's making some Darwin Award-level mistakes. Now a special category is being opened up to honor this digital absurdity.
Nominations are now open for the 2025 AI Darwin Awards, which organizers say will "honor those who use artificial intelligence in spectacularly reckless ways." Unlike the main Darwin Awards, the AI version will not focus on specific human tragedies. Instead, the spotlight falls on high-profile failures.
Here are some of the top nominees so far:
Taco Bell's Drive-Thru
Smith Collection/Gado
The fast food chain rolled out AI-powered ordering systems at more than 500 stores, then had to discontinue them when faced with a slew of complaints and customer trolling (such as the person who ordered “18,000 cups of water, please.”
OpenAI's GPT 5
NurPhoto
Just one hour after OpenAI rolled out the latest build of its popular chatbot (with much fanfare), researchers were able to jailbreak GPT-5, exposing tremendous security risks. The attack exploited the AI's reasoning capabilities, letting it unknowingly complete harmful requests without ever seeing any sort of malicious instructions.
Attempted fraud by an Airbnb host
In an attempt to scam users out of £12,000, a Superhost on Airbnb $ABNB used AI-created images to digitally manipulate photos to show damage to a table, mattress, appliances, and more. Problem was, the photos each showed different types and patterns of damage, alerting the company that the claim was bogus.
MyPillow lawyers
Helen H. Richardson/MediaNews Group/The Denver Post via Getty Images
As MyPillow CEO Mike Lindell was on trial for defamation, his attorneys leaned on AI to assist with their legal writing. The brief they submitted, however, featured nearly 30 defective citations, misquotes, and references to completely fictional court cases. Lindell was ultimately ordered to pay $2 million to Dominion $D Voting Systems.
McDonald's
Anna Barclay
The fast food chain had hoped the AI chatbot "Olivia" would be an asset that helped streamline the recruitment process. It spoke with, and collected personal information from, millions of candidates. What McDonald's $MCD failed to do, though, was give Olivia a robust security system. Instead it used the password 123456, which led to the exposure of information for 64 million job candidates.
Microsoft
Earlier this year, Microsoft $MSFT laid off 9,000 workers, a large number of which were in the Games division. An executive producer in that unit suggested impacted workers use AI tools like ChatGPT or Copilot to "help reduce the emotional and cognitive load that comes with job loss." The post was quickly deleted, but was still a huge PR black eye for the company.