The Center for Democracy and Technology’s 2026 wrap-up counts 146 chatbot, conversational-AI, and companion-bot bills proposed, passed, or enacted across the 2025-26 state sessions. Over the same window, the current Congress passed zero AI bills. The state beat has quietly consolidated around a single object of regulatory anxiety, and it isn’t frontier models.

Tech Policy Press pegs 2026 alone at more than 100 chatbot bills introduced and 14 enacted. The Transparency Coalition’s mid-August tally found 78 still alive in 27 states. The category that once dominated the discourse, algorithmic-discrimination frameworks modeled on Colorado’s SB 24-205, has receded: DOJ joined xAI’s April 24 suit against the Colorado law, and by May 14 the statute was repealed and replaced.

Connecticut SB 5, which CDT calls the most comprehensive 2026 state statute, bundles chatbot controls with a regulatory sandbox, ADS transparency, and whistleblower protections for frontier developers. Maine, Rhode Island, and Colorado went narrower and sharper, prohibiting AI therapy chatbots unless a licensed professional delivers them. New York transmitted a kids chatbot safety bill (S 9051) and an AI training-data transparency act to Gov. Hochul on June 1; she has until Dec. 31 to sign.

Federal activity was choreography, not law. A Senate committee advanced the GUARD Act (S. 3062) and NO FAKES Act (S. 4591); Sens. Curtis and Schiff introduced the SAFE KIDS Act (S. 4855). Trump pulled EO 14409 on May 21, then signed it June 2. NIST’s CAISI cut pre-deployment evaluation deals with Google DeepMind, Microsoft, and xAI on May 5.

That’s the picture: executive orders, evaluation MOUs, and litigation against the one state law that tried something structural. The states are legislating chatbots because Congress won’t legislate anything, and because chatbots are the AI product voters actually encounter.

Sources

Sources