States push AI rules
A widening list of U.S. states — including California and Utah — are moving ahead with their own AI guardrails despite federal calls to pause new rules, creating a compliance patchwork for multi-state businesses. That fragmentation contrasts with Europe’s more centralized regulatory approach, and experts warn the split could shape who wins commercially by favoring firms that can navigate divergent requirements. (nytimes.com) (fortune.com)
State legislatures have exploded with proposals: as of March 2026 lawmakers in 45 states had introduced roughly 1,561 AI-related bills, according to a multistate tracker. (multistate.ai) President Trump signed Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence,” on December 11, 2025; the order directs the attorney general to form an AI Litigation Task Force and asks federal agencies to evaluate and potentially condition federal assistance to states whose AI laws the administration deems inconsistent with the national policy. (federalregister.gov) California’s SB‑53 (the Transparency in Frontier Artificial Intelligence Act) was signed by Gov. Gavin Newsom on September 29, 2025 and took effect January 1, 2026, requiring frontier-model developers to publish safety frameworks and report incidents. (gov.ca.gov) SB‑53 targets the largest “frontier” models using an operations threshold specified in the law’s implementing discussions (often cited as around 10^26 integer or floating‑point operations across training and major fine‑tuning runs), and the statute includes civil penalties and whistleblower provisions that companies are parsing ahead of enforcement. (orrick.com) Utah’s AI regime began with S.B. 149 (the Artificial Intelligence Policy Act) enacted March 13, 2024 and effective May 1, 2024, and the state amended that framework in 2025 with bills (S.B. 226, S.B. 332, H.B. 452 and S.B. 271) that took effect May 7, 2025 to add disclosure rules, new privacy limits and a mental‑health‑chatbot statute. (gtlaw.com) Colorado’s broad AI statute (SB 24‑205) has been the subject of ongoing revisions and its operative date was delayed to June 30, 2026 after lawmakers failed to reach a compromise in a special session. (natlawreview.com) Congressional avenues to block state AI rules have faltered: lawmakers removed a provision from the NDAA that would have barred states from enforcing their own AI laws, leaving the state‑by‑state patchwork intact while legal fights play out. (fisherphillips.com) Survey and industry analyses show concrete economic pressure: a U.S. Chamber report found about 65% of small businesses worry the regulatory patchwork will raise litigation and compliance costs, and analysts project new categories of AI remediation and legal exposure could amount to billions in vendor and user costs by mid‑2026. (uschamber.com)
