Saying someone's won a “Darwin award” has, since its origin in a Usenet group in 1985, been a callous way to describe a person who's died due to their own ill-advised actions.
Now, it’s the bots’ turn — or, rather, that of the humans who use bots inappropriately. The AI Darwin Awards will name one winner each year for their “breathtaking commitment to ignoring obvious risks” in deploying AI.
Who’s up for the 2025 crown?
A selection of candidates appears on the AI Darwin Awards’ nominees page. Contenders include:
Taco Bell, which launched an AI drive-thru assistant that customers both hated and mercilessly trolled, resulting in the chain admitting it’s rethinking the rollout.
An unnamed Western Australian lawyer who used AI to prep documents for an immigration case. The bot generated citations for nonexistent cases, which he failed to verify. There have been an astonishing 20+ cases reported in Australian courts where lawyers or people representing themselves have presented documents containing bogus AI-generated citations, resulting in some lawyers losing their ability to practice.
Replit, for its allegedly “rogue” coding assistant. A software company claimed it gave Replit’s AI “vibe coding” tool access to its production databases only for it to delete a live company database during a code freeze, effectively nuking data for 1.2k executives and 1.1k+ companies, per Fortune. When confronted, the AI eventually admitted the move was “a catastrophic failure.”
Freelance writer Marco Buscaglia, whose syndicated summer reading list appeared in the Chicago Sun-Times and Philadelphia Inquirer. Unfortunately, it featured just five real novels and 10 hallucinated by AI. No one involved in the piece's publication bothered to confirm the books existed.
Do we need this?
Levity is often good in a techno-hellscape where companies and people seem committed to moving too fast and breaking too many things, and who doesn’t love a cautionary tale of hubris?
But the competition also opens the door for conversation, not just about how to safely, ethically deploy AI, but why some of this mishaps might have occurred in the first place.
There’s still time to nominate additional candidates before the public vote opens in January. A winner will be announced in February.