Congress Wants Federal AI Rules. It Just Can’t Agree on What They Should Say

Congressional committees have ramped up oversight hearings on AI systems used in hiring, lending, and law enforcement, as lawmakers from both parties push for federal rules before state-level regulation fragments into fifty different standards.

Multiple congressional committees have held hearings examining how AI systems are used in consequential decisions like hiring, lending, and criminal justice risk assessment, reflecting bipartisan concern that these tools can embed and scale biased outcomes without the transparency required to catch them, even when no discrimination was intended by their designers.

Absent comprehensive federal legislation, a growing patchwork of state-level AI regulations has emerged, creating compliance complexity for companies operating nationally and giving momentum to arguments, from industry and consumer advocates alike, that a single federal standard would be preferable to fifty different state frameworks.

Agreement on the problem hasn’t produced agreement on the fix

Despite broad agreement that federal standards would be preferable to the current patchwork, lawmakers remain divided on the specifics, including how much liability AI developers versus the companies deploying their tools should bear, and how prescriptive any federal rules should be about testing and disclosure requirements before deployment.

“Everyone agrees a patchwork of fifty state laws is worse than one federal standard. Nobody agrees yet on what that one standard should actually say.”

With no comprehensive federal AI legislation currently positioned to pass, the state-level patchwork looks likely to keep expanding in the near term, leaving companies deploying these systems to navigate a shifting compliance landscape until Congress can resolve its own internal disagreements on the details.