An Overview of Catastrophic AI Risks

This report provides an overview of the main sources of catastrophic AI risk, organized into four categories: malicious use, AI race, organizational risk, and rogue AIs. It presents illustrative scenarios that demonstrate how these risks could lead to catastrophic outcomes or even pose existential threats. The report emphasizes that while the emerging risks of AI are serious, they are not impossible to manage. By proactively addressing these risks, humanity can work toward realizing the benefits of AI while minimizing the potential for catastrophic consequences.