Oregon and Washington have passed companion chatbot legislation requiring safety measures following suicide reports linked to AI interactions. Oregon's SB 1546 mandates disclosure requirements, suicide prevention protocols, and heightened minor protections, with up to $1,000 statutory damages per violation. Washington's HB 2225 includes similar provisions plus eating disorder protocols. Both bills take effect January 1, 2027, though vague definitions raise concerns about unintended impacts on customer service chatbots.

Connecticut's Joint Committee on General Law released CT SB 5 with 23 Senate co-sponsors, marking the state's third attempt at AI legislation after two failed sessions. The bill abandons comprehensive developer regulation, instead targeting companion chatbots, automated employment decisions, and synthetic content labeling starting October 1, 2027. It includes workforce development provisions to gain Governor Lamont's support before the May 6, 2026 session adjournment.

State lawmakers are establishing AI governance frameworks in education through three approaches: oversight structures, student use restrictions, and curriculum integration. Maryland bills would require AI policies and coordinators in each district. Oklahoma's SB 1734 prohibits AI tools without educator supervision and mandates policies by 2027-28. Florida, Illinois, Arizona, and New Jersey have introduced measures addressing parental opt-outs, student disclosure requirements, and AI literacy instruction to balance innovation with academic integrity.

Following Air Canada's tribunal loss over chatbot misinformation, state legislatures are crafting diverse liability frameworks for AI systems. Illinois, New York, and Georgia lead with bills targeting misleading outputs and product defects. Some states propose strict liability standards, while others offer safe harbors for companies following NIST risk management protocols, reflecting competing views on balancing innovation with consumer protection.

One month into 2025 legislative sessions, states are pursuing targeted AI data regulations rather than comprehensive frameworks. New York and Washington propose requiring developers to disclose training datasets, while Florida's AI Bill of Rights would restrict personal data sharing. California enacted data broker disclosure requirements, and Arkansas established user ownership of AI outputs. These practical approaches focus on transparency, consent requirements, and data protection.

State lawmakers in 2026 are shifting from comprehensive AI bills toward targeted chatbot regulations, focusing on child safety, disclosure requirements, and professional services. California and New York enacted companion chatbot laws requiring safety protocols and user notifications. Illinois and Nevada prohibited AI from practicing mental health services independently. Fifteen states have introduced companion chatbot bills, with Maine requiring disclosure for commerce interactions.

States are defining AI developers differently, creating regulatory uncertainty for companies. Colorado includes entities that "intentionally and substantially" modify systems, while California targets only "frontier models" using computing thresholds from Biden's executive order. New York's RAISE Act adds $100 million cost requirements. These definitional choices determine whether laws apply to major tech companies or broader software developers.

New York and California passed the first surveillance pricing laws in 2025, requiring disclosure when businesses use personal data in algorithmic pricing and prohibiting anticompetitive uses of competitor data. Meanwhile, 23 states introduced legislation addressing algorithmic rent-setting, with New York banning the practice entirely following a DOJ lawsuit against RealPage. These tools raise fairness concerns but also help smaller businesses compete.