With Congress stalled, many states have moved ahead, passing dozens of laws to regulate artificial intelligence — focusing on child safety, corporate transparency and whistleblower protections. That flurry of state activity has raised tensions with the White House, which says a single federal framework is necessary to give innovators certainty and prevent a patchwork of conflicting rules.
In March the administration released a national AI legislative framework it wants lawmakers in Congress to adopt. Administration officials, including Michael Kratsios of the Office of Science and Technology Policy, have argued that only federal legislation can create the uniform legal environment developers need.
That federal push has at times altered state action. Utah Republican state Rep. Doug Fiefia, a former Google employee, said a transparency bill he sponsored never came to a vote after the White House circulated a brief memo opposing it; he says the memo called the measure unworkable without explaining why. A White House official speaking on background told NPR the administration has not told states they cannot pass child safety protections but declined to comment specifically on the memo described by Fiefia.
Many state lawmakers — including Republicans — say they feel they cannot wait for Congress. Fiefia and others argue that states are better positioned to move quickly on immediate harms like risks to children. Pennsylvania State Sen. Tracy Pennycuick, sponsor of the SAFECHAT Act aimed at preventing chatbots from encouraging self-harm or violence, said states can spot problems and respond faster than the federal government.
Some GOP officials acknowledge the risk of inconsistent rules but still favor state action until Congress acts. Texas State Sen. Angela Paxton warned against a confusing patchwork of laws, but said states should preserve their ability to legislate in the interim.
Reactions to the White House framework have been mixed. Supporters prefer one national standard over 50 different approaches, while critics say the framework is too vague and lacks strong accountability measures. Riki Parikh of the Alliance for Secure AI said the plan omits key issues such as job displacement and robust corporate accountability, and therefore should not replace work already underway in the states.
Tennessee Attorney General Jonathan Skrmetti called the framework a constructive step compared with an earlier White House proposal for a 10-year moratorium on state AI laws, a proposal that ultimately failed last year. Skrmetti nonetheless expressed concern about the administration’s ties to the AI industry.
Public polling reflects those worries: a January survey from Morning Consult and the Tech Oversight Project found a majority believe the administration is too close to Big Tech, while other polls show bipartisan support for AI regulation, sometimes with Republicans more inclined than Democrats to favor restrictions.
On Capitol Hill, some Republican allies back the White House principles but major legislation has not advanced. Sen. Marsha Blackburn is working with the administration on the TRUMP AMERICA AI Act, intended to codify the White House agenda. The administration says it is holding productive conversations with lawmakers.
For now, the struggle continues: the White House pressing Congress for a single national approach, and states — often led by Republicans — moving ahead with their own laws to address immediate harms and gaps.