According to TheRundownAI, California ordered a two‑month expert review to propose tougher AI safety rules including a kill switch and lab monitoring.