Bowen ZHOU (周伯文): Make Safe Al: At The Frontier of Agentic Al and Recursive Self-Improvement

In his keynote speech, Bowen ZHOU, Director and Chief Scientist of Shanghai AI Laboratory, introduced the distinction between “Make AI Safe” and “Make Safe AI,” highlighting the need to build AI systems with safety embedded throughout their evolution. Zhou discussed how rapid advances in frontier models, AI self-improvement, autonomous agents, and accelerated development cycles are challenging traditional approaches to AI safety. As AI capabilities advance faster than existing safeguards, he emphasized the need to move beyond one-time testing toward continuously verifiable and adaptive safety