Recent advancements in AI and LLM agents have unlocked powerful new capabilities across a wide range of applications. However, these advancements also bring significant risks that must be addressed. In this talk, Prof Dawn Song will explore the various risks associated with building and deploying AI and LLM agents and discuss approaches to mitigate them. Prof Dawn Song will also examine how frontier AI and LLM agents could be misused, particularly in cybersecurity attacks, and how they may reshape the cybersecurity landscape. Ensuring a safe AI future demands a sociotechnical approach. She will outline a recent proposal for a science and evidence-based AI policy, highlighting key priorities to deepen our understanding of AI risks, develop effective mitigation approaches, and guide the development of robust AI policies.