AI SAFETY WORKSHOP

Descrição

Come break a chatbot. We'll give you an AI assistant with safety rules, and your job is to make it ignore them. It's called prompt injection, it's one of the open problems in AI Safety, and by the end of the session you'll have done it yourself. But breaking things is only the start.We begin with an accessible introduction to what AI safety actually is: alignment, misuse risks, robustness with live polls and open discussion rather than a lecture you sit through passively. No coding, no maths, no prior knowledge assumed. Then we get hands-on. You'll experiment to try to defeat AI models' safeguards, and afterwards we'll unpack together why those vulnerabilities exist and what they mean for AI when it is deployed. Finally, we'll show you where to go next: SAIN Utrecht's course programme (AI Safety Fundamentals, AI Governance, and Technical AI Safety), conferences and events, fellowships, and the research and policy careers that this field has opened up. Attendees consent to filming and photography as participants in this event. By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please during the event let a member of our staff know so we can accommodate this.