The fifteen sort into three families of duty: disclosure, child safety, credential limits. They diverge sharply on enforcement — Oregon lets users sue for $1,000 a violation, while Idaho leaves the job to the attorney general.
Fifteen U.S. states now have chatbot statutes on the books, ten of them signed in 2026 alone, Bloomberg Law reports. One set of duties is already live: California's companion-chatbot law, SB 243, took effect Jan. 1, 2026.
Taken together, the fifteen do not add up to a single framework. They fall into three groups: rules requiring openness and notice to users, safeguards for children, and limits tied to professional credentialing. California, New Jersey and Maine require operators to tell the people using a chatbot that their conversation partner is artificial intelligence rather than a person, though the three statutes differ in their finer points. Nevada, Tennessee and Maine bar firms that supply AI from advertising or portraying their systems as credentialed mental health practitioners.
Of the three, the child-safety group is the largest and most detailed, and it expanded sharply in 2026, with California, Washington, Oregon, Colorado, Georgia, Iowa, Idaho, Nebraska, Hawaii and Rhode Island all enacting minor-protection frameworks. Those laws bar sexually explicit content, exchanges that play on emotions by imitating a romance, and erratic reward mechanisms built to keep young users coming back for more. They also require systems to spot signs of suicide and self-injury and to follow procedures for steering users toward crisis help.
How much a company stands to lose depends heavily on the state. California treats a system as covered where it can satisfy a user's need for social connection, leaves customer-service transactions outside its reach, and sets penalties of $5,000 for each violation on each day. Oregon's SB 1546 takes a different route, letting individuals bring suit themselves with damages fixed by statute at $1,000 for each violation; Oregon is among a handful of states whose private-suit provisions markedly raise the financial risk operators face. Idaho's S 1297 leaves enforcement to the attorney general, and private suits are not permitted.
Which systems fall inside those rules is itself unsettled. Definitions of companion or conversational AI differ from one state to the next, leaving it unclear how far any particular law reaches. Oregon's is drawn more tightly and built around conduct: it covers AI companions that mimic an ongoing, humanlike bond and hold on to context so exchanges can be tailored to the user, while leaving out the ordinary chatbots that answer customer questions on limited, self-contained topics.
Most of the duties in these statutes, Oregon's included, have yet to take hold. Colorado's HB 1263, enacted May 29, 2026, takes effect Aug. 12, 2026; it requires notification and parental access tools, bars sexually explicit or emotional-dependence content, and rules out gamifying engagement for anyone under 18. Oregon's SB 1546 follows on Jan. 1, 2027, and Nebraska's Conversational Artificial Intelligence Safety Act and Idaho's statute on July 1, 2027.
All of it sits under a federal question. Trump administration officials have objected loudly to AI rules written at the state level, and a measure with backing from both parties that would override state chatbot statutes has come up for discussion in Congress. New York legislators have campaigned in the other direction, pressing lawmakers in Washington to resist any such override.