This project explores how policymakers can respond to the challenges and opportunities posed by frontier AI systems – highly capable models with potentially transformative and risky capabilities. Our work includes a deliberative policymaking framework that complements existing risk-based regulation by focusing on freedom, understood as both the expansion of human capabilities and the creation of opportunities. A further focus is on understanding and managing instrumental goals – the self-directed or power-seeking behaviours advanced AI systems may develop as they pursue complex objectives. Rather than treating these behaviours solely as failures, the project explores how they can be anticipated, monitored and redirected through appropriate governance structures.