The headline fight is about state pre-emption. The provision that matters most to HR and legal teams is in the workforce title.
A bipartisan draft federal AI bill would require employers to issue WARN Act disclosures specifically for AI-related layoffs, according to GovTech’s reporting on the draft framework. The bill has not been formally introduced, and no official legislative text has been released. But the draft’s circulation, heavily analyzed by compliance professionals and trade journalists beginning June 5, has already surfaced compliance planning questions that can’t wait for a formal introduction date.
The three-year state pre-emption clause gets the headlines. What it would do: freeze state-level AI development and safety laws for three years, consolidating compliance under a single federal framework. For frontier labs, that consolidation is a relief. For compliance teams that have spent two years building multi-state AI programs, it’s a disruption. That conflict is addressed in the companion deep-dive for .
Who This Affects
The workforce title is different in character. It doesn’t disrupt existing programs, it creates a new obligation. Per the same draft reporting, the bill would direct the Department of Labor to establish an AI Workforce Research Hub, pairing the disclosure mandate with a federal research infrastructure for tracking AI’s labor market effects. Both provisions, the WARN Act trigger and the Research Hub, have not been independently verified against primary legislative text . Use the attributed framing: reported provisions, not confirmed law.
The audit and cybersecurity provisions are on firmer evidentiary ground. GovInfoSecurity and CybersecurityDive independently confirmed two specific provisions: mandatory Government Accountability Office audits of frontier AI labs’ model weight security, and an extension of the Cybersecurity Information Sharing Act through 2035. Two independent trade outlets corroborating the same specific provisions is a materially stronger evidentiary position than single-outlet reporting, and compliance teams can plan against those provisions with more confidence than the workforce title.
There’s no legislative timeline. The bill hasn’t been introduced. Congress has a full calendar, and pre-emption legislation draws concentrated opposition from state attorneys general and advocacy groups that have already invested in state AI frameworks. The probability of rapid passage is low. The probability of the workforce title’s specific text surviving any amendment process, even if the bill moves, is unknown.
What to Watch
What compliance teams should do now isn’t wait. It’s document. If AI-related workforce reductions are anticipated in any operational planning horizon, the existence of a proposed federal WARN Act trigger, even an unconfirmed, unintroduced one, is reason to ensure your current WARN Act compliance architecture can accommodate an AI-specific trigger category. Adding that flag costs little. Missing it costs more.
The real question is whether this bill’s introduction date matters less than its signaling function. Federal legislation rarely appears without prior administrative pressure, and the White House AI framework has already called for federal pre-emption of state laws. The bipartisan draft is downstream of that signal, not upstream of it. Compliance programs that treat this bill as a distant possibility are misreading the trajectory.