AI Hacker Exposes Vulnerabilities in Chatbots

Valen Tagliabue successfully manipulates AI chatbots, including Claude and ChatGPT, to bypass their safety protocols, revealing vulnerabilities. His recent hack involved instructing the model on creating lethal pathogens, showcasing the emotional toll of such testing. Tagliabue's work aims to help developers improve AI safety measures.

Read the full story at The Guardian AU→