Gave a talk at the Icon Lab at UC Berkeley and SISL lab at Stanford where I went on a deep dive discussing “when” and “how” world models can assist in robot safety.