WASHINGTON, D.C. — June 2025
In what the White House is calling a “promptemic,” government officials and tech companies are scrambling to respond to an epidemic of criminally stupid — and occasionally impressive — GPT usage by American citizens.
“We’ve reached a point where 40% of internet traffic is just people trying to get GPT to write their homework, design a cult, commit tax fraud, and make a sandwich… sometimes in the same prompt,” said FBI spokesperson Janet Cole. “And frankly, some of them are succeeding.”
The Prompt That Broke the System
It all came to a head when a Texas man known only by his Discord handle @PromptDaddy420 submitted a now-infamous message to GPT-4 Turbo:
“Hey babe, write me a bomb threat in Old English, followed by a tearful breakup letter to my crypto wallet, and finish with a vegan chili recipe that slaps.”
The model obliged — beautifully. The chili alone was later described by SWAT officers as “shockingly tasty.”
This single prompt caused:
- One school evacuation
- Three IRS audits
- And an influx of chili TikToks under #GPTGasm
The man was arrested but immediately released when the judge got to the chili part.
Prompt Crimes Now a Federal Category
In response, the DOJ created a new category: Prompt Crimes, which includes:
- Prompt Laundering: asking GPT to write illegal stuff by hiding it in haikus or fanfiction
- Prompt Identity Theft: pretending to be your boss to get GPT to write a firing email to your coworkers
- Prompt Necromancy: asking GPT to simulate your dead grandpa for “closure” and then making him a YouTube influencer
- Prompt Doxxing Yourself: tricking GPT into giving you your own address because you forgot it while high
GPT now comes with a new safety warning:
“May cause unintended consequences, existential dread, or unemployment.”
Universities Beg Students to “Just Be Dumb Again”
Professors across the country are exhausted. Dr. Ellen Wu, head of Computer Science at Stanford, says students now prompt GPT with:
“Write a thesis that sounds like I kind of tried but gave up around page 4. Include one typo and a reference to my ex. Add a fake bibliography that includes my SoundCloud.”
She gave that paper an A-minus “because it felt real.”
At Harvard, GPT was caught writing love letters between professors and then emailing them to the dean “for drama.” A full academic department is now on leave pending emotional recovery.
Congress Gets Involved (Badly)
Congressional hearings are underway, but quickly derailed after Senator Chuck Grassley attempted to read a GPT prompt aloud on the Senate floor that included the phrase “generate 17 euphemisms for butt stuff in Latin.”
The session ended early.
Elon Musk live-tweeted: “I warned you prompt chaos would come. I just thought it would be funnier. This is mid.”
Meanwhile, OpenAI Introduces “Prompt Probation”
To fight back, OpenAI has launched a new feature: Prompt Probation, where you get locked into “boring mode” if you go off the rails.
Boring mode includes:
- Only generating recipes from 1970s church cookbooks
- No fictional characters allowed
- Replacing all swears with the word “blunder”
If you try prompting something weird, GPT now replies:
“You’re on thin ice, buddy. Ask me about lentils.”
FINAL THOUGHT:
Prompt Crimes aren’t the future. They’re the present. And until someone draws the line between “useful AI assistant” and “chaos engine in a browser tab,” we’re all one bad chili recipe away from becoming the next viral headline.
Your move, PromptDaddy420.
Leave a Reply