The Center for Democracy and Technology counted 146 bills targeting chatbots, conversational AI, or AI companions moving through state legislatures across the 2025–2026 sessions, according to a legislative update it published August 20. Chatbots, not frontier models, are what statehouses actually regulated.

The broader tally is smaller but still growing. Tech Policy Press reports 29 states had enacted AI legislation as of July 1, 2026, off 2025’s pace but steadily thickening the patchwork Washington keeps promising to preempt. California is expected to add more before year-end.

CDT flagged Connecticut SB 5 as the session’s most comprehensive law, bundling a regulatory sandbox, chatbot controls, an auditor study, and ADS transparency and anti-discrimination requirements into a single vehicle. California SB 243, effective January 1, 2026, forces companion-chatbot operators to disclose non-human status and add minor protections. Colorado went the other way, using SB 189 to scale back its 2024 algorithmic discrimination law after constitutional challenges, per White & Case.

The chatbot fixation has a darker perimeter. 42 states now criminalize AI-generated CSAM after Alaska, Idaho, Michigan, and Florida joined in 2026. Minnesota (HF 1606) and Texas (SB 441) enacted nudification statutes; xAI is suing Minnesota, arguing HF 1606 is unconstitutional.

Federal action has been almost entirely executive. Trump signed Executive Order 14409 on June 2, 2026, directing classified benchmarking of “covered frontier models” while explicitly barring a mandatory federal licensing regime. A December 2025 order created an AI Litigation Task Force to challenge state laws judged not “minimally burdensome” and told Commerce to study withholding BEAD broadband funding from states with “onerous” AI regimes.

Congress passed no new AI statutes. The regulator of American AI in 2026 is a coalition of state attorneys general and a White House trying to sue them out of the job.

Sources

Sources