Adam Gleave

Adam Gleave is the CEO of FAR.AI, a non-profit research institute dedicated to making advanced AI systems trustworthy and secure. FAR.AI’s red-team conducts pre-deployment testing for developers including OpenAI and post-deployment testing on behalf of governments including the EU AI Office, and has discovered universal jailbreaks in models from all frontier developers. Adam’s own research has identified scaling laws for robustness, and found adversarial examples in robotics and superhuman Go AIs. Prior to founding FAR.AI, Adam completed his PhD in AI at UC Berkeley and briefly worked at Google DeepMind. Outside of FAR.AI, Adam is an expert on the EU AI Act’s Scientific Panel, and a board member of METR, LISA and SAIF.