Beyond the Model Boundary: A Systems Perspective on AI Safety

AI safety is not only about building better models; it is also about understanding what happens when AI becomes part of real-world systems. This lightning talk explains how frontier AI can reshape risk as it is used by people, connected to tools, embedded in organisations, and linked through complex dependencies, changing how problems arise, spread, become harder to control, and how it tests our ability to respond. The key takeaway is simple: safer AI requires safer systems, not just safer models.