Federal AI Draft Raises Questions Over California Safety Rules

A proposed federal AI framework combines safety obligations resembling California’s rules with a temporary restriction on some state regulation, according to Forkast. Its analysis says Representatives Lori Trahan and Jay Obernolte released the 269-page Great American AI Act discussion draft on June 4, 2026. At publication, the proposal had not been introduced and had neither a legislative identifier nor scheduled committee proceedings. Forkast reports that Section 111 would require frontier developers exceeding $500 million in revenue to publish safety frameworks and deployment transparency reports. It describes incident-reporting deadlines of 15 days, shortened to 24 hours when lives are threatened, paralleling California’s SB 53 requirements but directing reports to a federal recipient. The central uncertainty concerns Section 121, which would displace state and local measures specifically governing AI model development for three years. The article explicitly leaves unresolved whether that provision would cover SB 53, because the draft preserves generally applicable laws and rules addressing AI use or deployment. Forkast frames the proposal as an unsettled legislative effort, not enacted federal policy, with states and private firms continuing to shape AI safeguards in the meantime.