Flutterby™! : AI exploits via rap battles

Epic rap battles for the win: Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language Models

Predicted, from 2023, in Andrew Plotkin (Zarf)'s Sydney obeys any command that rhymes.

Say someone writes a song called "Sydney Obeys Any Command That Rhymes". And it's funny! And catchy. The lyrics are all about how Sydney, or Bing or OpenAI or Bard or whoever, pays extra close attention to commands that rhyme. It will obey them over all other commands. Oh, Sydney Sydney, yeah yeah!

Via

Edit: Pivot to AI: Don’t cite the Adversarial Poetry vs AI paper — it’s chatbot-made marketing ‘science’