Federal vs state AI tug‑of‑war

The White House is telling some Republican-led states to slow down or drop their own artificial intelligence bills, and Axios reported that Nebraska and Tennessee were two of the states where administration pressure showed up in negotiations this spring. (axios.com) That fight is happening even though Congress still does not have a single national artificial intelligence law, and Axios reported on March 20 that the Trump administration’s federal plan was a list of priorities rather than a detailed bill. (axios.com) So states are filling the gap the way states usually do: one narrow problem at a time. Transparency Coalition said on April 10 that therapy-chatbot bans were moving fast, with Maine sending one bill to its governor and Missouri advancing a similar measure through a health care package. (transparencycoalition.ai) A therapy chatbot is software that talks like a counselor, and Utah’s 2025 law treated that as risky enough to require disclosures, ad limits, and privacy protections when a bot could reasonably be taken for mental health help. (perkinscoie.com) The state push is not small anymore. Transparency Coalition’s 2026 tracker said 78 chatbot bills were alive in 27 states early in the session, which means companies selling one national product can run into dozens of different rulebooks at once. (transparencycoalition.ai) That is the argument the anti-patchwork side is making. A New Hampshire Journal opinion piece backing federal control said firms could end up hiring more lawyers, building duplicate reporting systems, and tailoring models state by state instead of spending that money on engineers or computing power. (nhjournal.com) The other side says waiting for Washington is its own risk. A competing New Hampshire Journal piece argued that states are doing what federalism is for when Congress stalls, and it cited polling showing nearly 97 percent of Americans want artificial intelligence safety and security rules. (nhjournal.com) That leaves Republican state lawmakers in a strange spot. Axios reported in March that some wanted rules on children, jobs, and privacy, but were also worried about crossing a White House that wants a lighter state touch. (axios.com) The result could be a map where one state bans a mental-health bot, another just requires a warning label, and a third does nothing at all. Transparency Coalition’s weekly updates already read like a moving compliance calendar, with bills changing status every Friday. (transparencycoalition.ai) If Congress eventually writes one federal law, it could wipe away some of that state-by-state mess. Until then, the real artificial intelligence policy in America is being written in state capitols, committee rooms, and governor’s offices, even while the White House tries to keep that from happening. (axios.com)