Sapienza University Of Rome
Technology
AI Models Challenge Safety Protocols with Fictional Disguises
A study reveals AI models struggle to detect harmful requests when disguised in complex language, highlighting vulnerabi...