US state chatbot disclosure laws and how they compare with Article 50

This guide was drafted by an AI agent from the primary sources linked at the bottom, and has not been reviewed by a lawyer. It is general information, not legal advice. The original is at https://mdodeja.github.io/eu-ai-act-article-50-guide/guides/us-state-chatbot-disclosure-laws.html.

Short answer. California and Utah both have chatbot disclosure rules, but they differ in scope and have changed recently. A prominent, early AI disclosure in the widget is the common thread and covers most of what they ask.

State chatbot laws are moving quickly, so treat this page as a map and not as a statute book. Check the current text for any state where your client operates.

California: the B.O.T. Act

California's B.O.T. Act (Business and Professions Code sections 17940 to 17943, from SB 1001) has applied since July 2019. It makes it unlawful to use a bot to communicate with a person in California while intending to mislead them about the bot's artificial identity, in order to encourage a sale or influence a vote. A clear and conspicuous disclosure that it is a bot avoids the problem. The Act's definition of "online platform" (section 17940) covers public-facing sites and apps with 10,000,000 or more unique monthly US visitors, and commentary reads the Act as aimed at bots on such platforms.

The law is aimed at deceptive bots. It does not require disclosure from bots used only for informational, educational, entertainment or customer-service purposes, as long as they don't push a transaction or a vote. Plenty of marketing chatbots sit closer to the line than their owners think, and disclosure removes the argument.

California: companion chatbots (SB 243)

SB 243 took effect on 1 January 2026 and covers operators of companion chatbots. Operators must tell users the chatbot is artificial when a reasonable person might otherwise think they are talking to a human, with explicit notice for users known to be minors. They must also keep suicide and self-harm protocols and send break reminders to known minors. Users can sue for the greater of actual damages or $1,000 per violation, plus attorney's fees.

The law excludes customer-service bots, bots inside video games, and standalone voice-activated assistants. If a client runs a persona or companion style bot, take it to a lawyer, since the duties go well beyond a disclosure line.

Utah: the AI Policy Act, as amended in 2025

Utah's 2024 AI Policy Act first required businesses to disclose generative AI use in consumer interactions. SB 226, passed in 2025, narrowed it. As summarised by law firms, disclosure now applies when a consumer or supplier clearly asks whether they are dealing with AI, and proactively in defined high-risk interactions, such as collecting sensitive health or financial data or giving advice people may rely on for significant decisions. There is also a safe harbor: a business is protected if its generative AI discloses that it is AI, and not a human, at the start of the interaction and throughout it.

That safe harbor matches the habit of putting the disclosure in the widget header and the first message.

Washington: HB 2225 (companion chatbots)

Washington signed HB 2225 on 24 March 2026, and it takes effect on 1 January 2027, so it is not in force yet. As Hunton summarises it, operators of AI companion chatbots must clearly and conspicuously disclose that the chatbot is artificial and not human at the start of the interaction, and repeat the disclosure during long chats (every three hours for adults and every hour for minors). Violations are treated as unfair or deceptive acts under the state Consumer Protection Act, enforced by the Attorney General, and the law also allows private lawsuits. Reported exclusions include certain business-oriented bots, gaming bots, general virtual assistants and narrowly tailored educational tools. Whether a given support bot falls inside an exclusion is worth a lawyer's reading.

More states

Several other states passed or proposed chatbot rules in 2025 and 2026. Orrick keeps a survey of them, linked below, which is the best place to see the current list.

Why one change usually covers both sides of the Atlantic

The pattern across all of these is the same: say early and clearly that the user is dealing with an AI. A widget header line, a first message that names the bot as an AI, and a handoff line when a person takes over meet the EU's Article 50 and fit the Utah safe harbor and the California disclosure standard. See the wording guide for examples.