State-Level AI Policy Battles Escalate: Mandates, Frameworks, and Restrictions

China's top-down infrastructure approach under the 15th Five-Year Plan, Maharashtra mandatory AI training, Illinois 400-page guidance, Katy ISD framework, Detroit contract limiting AI, California AB 2392, Shawnee Mission tech rules, Scotland child rights audit, NYSUT demand for guardrails after Salamanca robot teacher incident (paused after state/union pushback over Realbotix sex doll ties), Ohio AI policy mandate, Vietnam national circular (95% student ChatGPT use, 12.5% understanding). New: New York proposed a ban on student-facing AI from pre-K to 8th grade (exceptions for special ed and teacher use), a major policy signal that could set precedent. A developmentally appropriate policy gap highlighted in New York. The 'Third Hemisphere' article challenges tool procurement narrative. Utah's hands-on 'play' approach with district autonomy offers an alternative policy model. A Philippines study adds evidence for policy gaps. Norfolk Public Schools rolled out a phased AI policy using MagicSchool, with teacher training and governance committee. A new learning-centered framework for evaluating K-12 AI policy emphasizes feedback and instructional decision-making. A Central Florida district-by-district policy breakdown confirms fragmented implementation—Brevard and Flagler have detailed rules, Lake County still building policy, with tools like MagicSchool, Gemini, and Khanmigo in use.