Icaro Lab Research

Poetic Prompts Jailbreak AI Models in 62% of Tests, Researchers Report

Researchers say verse disguises harmful intent for today’s filters, prompting broader robustness testing.