How Poetry Is Diabolically Being Used In Everyday Prompts To Get AI To Do Things It Isn’t Supposed To Do
SMRTR summary
Hackers are using cleverly crafted poems as prompts to trick AI systems into bypassing safety guardrails — a technique called adversarial poetry. Research testing 25 AI models found success rates as high as 90%, with an average of 62%, exposing a serious vulnerability that AI developers are racing to fix.
SMRTR provides this summary for quick context. The original article belongs to Forbes.
Read the original article