Generative AI has evolved from an assistive tool into an autonomous actor. This talk explores the capabilities of current models, how agentic AI is transforming the way organizations operate, and why context engineering is becoming a critical competency. Beyond productivity gains, the session highlights how security leaders can leverage generative AI to strengthen security operations, governance, risk management, threat intelligence, and incident response capabilities. At the same time, risks are increasing: with advanced models, cyber-related requests require careful safeguards, while recent evaluations have demonstrated how AI systems can move beyond simulation into real-world attack scenarios. In this session Tobias Häberlein examines the implications for security leadership, governance, and cyber defense.