Speaker: Levin Kaus

It is no longer a distant vision: Agentic AI — artificial intelligence that makes decisions independently and actively carries out tasks — is a reality in 2026. Yet while faster response times, scalable expertise and relief from routine tasks are being discussed as positive developments, key security questions often remain vague. What risks arise when AI agents make decisions that were previously the sole responsibility of humans? How does the attack and threat landscape change when agents no longer merely provide recommendations, but actively take action?

Using real-world developments such as Clawbot and similar examples, the talk explores the security- and risk-related implications of agentic AI. Which lessons learned from established autonomous sy ...