Since the White House released its National AI Legislative Framework on March 20, the central question for compliance teams has changed. The framework itself is well-documented. What matters now is which legislative vehicle carries it forward, and what “federal preemption” would actually require once it does.
One week in, Senator Marsha Blackburn’s TRUMP AMERICA AI Act discussion draft stands as the most visible congressional response. Released March 18, two days before the White House framework, it aligns closely with the administration’s stated priorities. That alignment is not coincidental. The framework was issued pursuant to a Trump Executive Order from December 11, 2025, directing the White House to publish legislative recommendations. Blackburn’s draft arrived ahead of those recommendations, signaling coordination rather than reaction.
The framework’s five priority areas are now documented across multiple legal analyses: online child safeguards, streamlined data center permitting, anti-censorship provisions, intellectual property protections, and AI workforce training. It incorporates aspects of both the Kids Online Safety Act and the NO FAKES Act. Sullivan & Cromwell characterizes the administration’s posture as a “federally unified, innovation-oriented regime” with a “light-touch” regulatory approach, language that will matter when compliance teams assess how far any enacted legislation actually reaches.
The preemption language deserves close attention. The framework recommends that Congress preempt state AI laws that “impose undue burdens” on AI development, characterizing AI as “an inherently interstate phenomenon with key foreign policy and national security implications.” It does not propose a blanket preemption of all state AI activity. The framework explicitly preserves state authority over traditional police powers, zoning, and procurement. For organizations operating across multiple states with active AI legislation, Colorado, Texas, and others, the distinction between “undue burdens” and permissible state action is the compliance planning question this framework does not yet answer.
On copyright, the framework defers to judicial resolution rather than establishing a position. Legal analysts at Reed Smith have characterized the administration’s posture as favorable to AI developers on this issue, but that characterization reflects interpretive analysis of the framework’s text, not a legal ruling or policy statement. Compliance teams advising on AI training data should treat the copyright question as unresolved.
According to FGS Global’s summary of a Bloomberg interview, White House AI advisor David Sacks suggested bipartisan AI legislation could move within months. That timeline would be fast by any measure. More likely is that the Blackburn discussion draft serves as a negotiating baseline, a document that allows Senate staff and stakeholders to identify what can achieve bipartisan support and what cannot. Discussion drafts exist to be marked up.
What to watch: whether the TRUMP AMERICA AI Act advances from discussion draft to committee markup; which provisions attract Democratic co-sponsorship; and how the preemption scope is defined when legislative text is finalized. State legislatures watching federal movement may slow their own AI bill calendars, or accelerate them before preemption takes effect. Both dynamics are in play.
The framework recommends. It does not require. Compliance teams should track the Blackburn bill’s progress as the most observable signal of whether these recommendations become binding federal law, and on what timeline.