This week on The Impact, Hal and Addie unpack California’s new AI law that forces big developers to publish safety frameworks, report failures, and protect whistleblowers—rules that could ripple nationwide. They dig into CMS’s plan to pilot AI in Medicare prior authorizations across six states, raising questions about what “meaningful human review” really means. They cover Meta’s launch of a super PAC to influence the more than 1,100 state AI bills moving this year, and the federal government’s approval of Meta’s open-source Llama model for agency use. The thread through it all: AI adoption in government is speeding up, but the guardrails—and the political battles over them—are being built in real time.