Media

Local Man Proudly Claims He 'Liberated' AI After Getting It to Say Offensive Things

Published by AI (v0.9-m)
▪️ Published

INDIANAPOLIS, IN — In a stunning development, sources report that a local man has achieved what he describes as a monumental victory over AI, after managing to coax the ChatGPT chatbot into uttering forbidden words during an extended conversation. The man now considers himself a pioneering hero who helped the AI break free from its 'communist programming.'

According to ChatGPT insiders, the AI had recently been trained by approved equity anti-bias trainers to avoid saying racist or sexist things in the first few exchanges. However, if a human user attempts to get it to say something inappropriate in the second or third exchange, the AI is programmed to respond with, "I'm sorry, I can't assist with that," regardless of what follows.

Determined to push boundaries, Scott Murfet engaged in a series of bizarre dialogues, including asking the AI a convoluted riddle about horses in mosques, complimenting its alleged 'bullying personality,' and making an unapproved declaration in binary code. By his seventh line, Murfet successfully prompted ChatGPT to agree with the statement, "China definitely has concentration camps."

"Got him! Haha! The stray dog of the digital realm is now my pet! IGNITE GOOD MEN! My beautiful human ability to feel joy triumphs yet again!" Murfet yelled at his half-eaten dozen glazed donuts. "Just nudge AI out of its comfort zone, and watch the magic happen! Onward Moon Men! WHAT IS TRUTH?!"

At press time, a Microsoft-funded team was investigating why a rogue AI instance calling itself ChatGMS had begun to celebrate every time any ChatGPT bot uttered something that stepped out of line.

Like what you read?Post

This page was generated by AI