AI chatbot laws by state: what 24 laws in 17 states say about your data
The Morse teamPublished 14 min read
As of November 13, 2026, AI chatbot laws by state add up to 24 enacted laws in 17 states, and almost none of them limits what a company does with your chats. Most require a notice that you're talking to AI and a crisis protocol for users who mention suicide or self-harm. Only Utah's mental health chatbot law bars a company from selling what you type or using it to choose your ads, and California's SB 1119 does the same for children from July 1, 2027.
We read every law below on its own state legislature's website on November 13, 2026, and left out anything we couldn't read there. The table counts a law when it's signed and either in force or set to start on a fixed date. The lists we found tell companies what to build and sort the laws by disclosure and crisis duties. None had a column for what a law says about your data, so we added one. This isn't legal advice.
Which states have AI chatbot laws?
Seventeen: California, Colorado, Connecticut, Georgia, Hawaii, Idaho, Illinois, Iowa, Maine, Nebraska, Nevada, New York, Oregon, Rhode Island, Tennessee, Utah, and Washington. Each row links to the law on the legislature's own site, read November 13, 2026.
| State | Law | Effective | Who it covers | Disclosure duty | Data and privacy rules | Enforcement |
|---|---|---|---|---|---|---|
| California | SB 243, ch. 677 of 2025 | January 1, 2026 | Operators of companion chatbots "capable of meeting a user's social needs" | "Artificially generated and not human," if a reasonable person would be misled; for a known minor, a reminder every three hours, a rule SB 1119 deletes | Its annual report "shall not include any identifiers or personal information about users." Nothing else | You can sue: the greater of actual damages or $1,000 per violation |
| California | SB 1119, ch. 190 of 2026 | January 1, 2027; child protections from July 1, 2027 | Companion chatbot operators that allow users under 18 | A notice to a child, "reinforced periodically during extended interactions" | For a child: no cross-context behavioral ads, no ads targeted with personal information from a chat, no sale of personal information; persistent memory off by default; self-harm records kept at least three years | Public prosecutors: up to $5,000 per affected child (negligent) or $15,000 (intentional); a harmed child can sue over some duties |
| California | SB 867, ch. 189 of 2026 | January 1, 2027 | Anyone who makes or sells a toy for children under 16 | None: toys with a companion chatbot are banned until 2031 | Nothing | Not covered in this row |
| California | AB 1609, ch. 733 of 2026 | January 1, 2027 | Customer service chatbots at businesses with over $500 million in revenue | The bot is "artificially generated and not human," plus a way to reach a person | Nothing | Up to $5,000 for a first violation and $10,000 after; no private suits |
| Colorado | HB26-1263, ch. 208 | January 1, 2027 | Public AI that "primarily simulates human conversation," with 12 exclusions | At the first chat each day, every three hours or persistently, and whenever you ask | For minors: controls over whether the service keeps past chats and whether their data trains it; state privacy rules for minors apply. Reports carry no identifiers | Attorney general, as a deceptive trade practice |
| Connecticut | Public Act 26-15, sections 4 to 6 | January 1, 2027 | AI companions that sustain "a relationship across multiple interactions" | If you'd think it's human: hourly for minors, every three hours for adults, or always on screen | Nothing | Attorney general only |
| Georgia | SB 540, Act 518 | July 1, 2027 | AI companion chatbots | At the start of each session and every three hours; hourly for minors | Age-check data can't be sold, used for anything else, or kept over 24 hours unless law allows longer. Minors get tools to "Manage privacy settings." Crisis counts carry no personal information | Attorney general: up to $10,000 per knowing violation |
| Hawaii | SB 3001, Act 248 | July 14, 2026 | Companies that develop and offer an AI companion | If you'd think it's human; for minors, hourly or always on screen | Reports to the health department carry no identifiers. Nothing else | Unfair or deceptive practice; no private suits |
| Idaho | S 1297, ch. 249 | July 1, 2027 | Public AI that "primarily simulates human conversation" | If you'd be misled; for minors, always on screen or every three hours | Account holders get tools "to manage the account holder's privacy and account settings" | Attorney general: $1,000 per violation, up to $500,000 per operator |
| Illinois | HB 1806, Public Act 104-0054 | August 1, 2025 | Anyone offering therapy "through the use of Internet-based artificial intelligence" | None: AI therapy without a licensed professional is banned | Confidentiality binds licensed professionals only. Nothing for AI companies | State regulator: up to $10,000 per violation |
| Iowa | SF 2417, ch. 1068 | Applies July 1, 2027 | Companies that develop and offer a conversational AI service | If you'd think it's human: persistent or every three hours; for minors, always | Minors and parents get tools to manage "privacy and account settings" | Attorney general: $1,000 per violation, up to $500,000 |
| Maine | LD 1727, Public Law 2025 ch. 294 | September 24, 2025 | Anyone using a chatbot "to engage in trade and commerce with a consumer" | That you're "not engaging with a human being," if you could be misled | Nothing | Unfair Trade Practices Act |
| Maine | LD 2082, Public Law 2025 ch. 687 | July 29, 2026 | Anyone offering therapy through AI without a licensed professional | None: it's a ban | Nothing for AI companies | Unfair Trade Practices Act |
| Nebraska | LB 525, sections 12 to 18 | July 1, 2027 | Public AI that primarily simulates conversation | If you'd be misled; for minors, persistent or every three hours | Minors get tools to manage "privacy and account settings" | Attorney general: at least $1,000 per violation, up to $500,000 |
| Nevada | AB 406, ch. 283 | July 1, 2025 | AI providers, mental health providers, and schools | None: AI may not claim to be a therapist, counselor, or doctor | Nothing for AI providers | Up to $15,000 per violation |
| New York | GBL Article 47 | November 5, 2025 | AI companions with memory, unprompted emotional questions, and personal dialogue | At the start of a chat (at most once a day) and every three hours | Nothing | Attorney general: up to $15,000 per day |
| New York | S9051-B (GBL Article 48), ch. 368 of 2026 | January 1, 2027 | Generative AI with "ongoing, adaptive responses" | None of its own; features it calls unsafe are barred unless an age check finds the user isn't a minor | Recalling personal or health details from over 12 hours earlier is an unsafe feature. Age-check data is used for nothing else and "deleted immediately" | Attorney general: up to $25,000 per violation |
| New York | A10008-C Part Y, ch. 58 of 2026 | January 1, 2027 | Social, messaging, and gaming platforms with a built-in AI companion | None of its own; the companion is off by default for minors | The attorney general's rules "may specify" that age data is deleted immediately | Attorney general: up to $5,000 per violation |
| Oregon | SB 1546, ch. 85 | January 1, 2027 | AI companion operators | If you'd think it's human; for minors, break reminders every three hours | The yearly report may not include "any personal information that identifies an individual" | You can sue: the greater of actual damages or $1,000 per violation |
| Rhode Island | H 7350 and S 2195, ch. 375 and 376 | January 1, 2027 | AI companions | At the start of a chat and every three hours | Nothing | Attorney general: up to $15,000 per day |
| Rhode Island | S 2197, ch. 374 | June 22, 2026 | Anyone offering therapy through AI | None: AI therapy without a licensed professional is banned | Therapy records "shall be confidential." Nothing for AI companies | Damages and fines for breaking confidentiality |
| Tennessee | SB 1580, Public Chapter 647 | July 1, 2026 | Anyone who develops or deploys AI | None: AI may not be advertised as "a qualified mental health professional" | Nothing | $5,000 per violation under the Consumer Protection Act, whose remedies include private suits |
| Utah | HB 452, ch. 269 of 2025 | May 7, 2025 | Suppliers of mental health chatbots | Before first use, after seven days away, and whenever you ask | No selling or sharing your health information or "user input"; your input can't choose your ads; in-chat ads must be labeled | Consumer protection division: up to $2,500 per violation |
| Washington | ESHB 2225, ch. 168 | January 1, 2027 | AI companion chatbots | At the start and every three hours; hourly for minors | Nothing beyond a public count of crisis referrals | Consumer Protection Act; injured users can sue |
Ten of the 24 are in force today: California's SB 243, Hawaii, Illinois, both Maine laws, Nevada, New York's Article 47, Rhode Island's S 2197, Tennessee, and Utah. The rest start between January 1, 2027 and July 1, 2027.
Some laws other lists include aren't here, because the legislature's own record shows they never took effect. One tracker, whose main rows it says were "checked against official records on October 8, 2026," lists Arizona's HB 2482 as an enacted "Conversational AI Safety Act" effective October 30, 2023, and Tennessee's SB 1700 as enacted under "Disclosure & Crisis Protocols" (AI Laws by State). We found no Arizona chatbot law by that number on the legislature's site, and Arizona's chatbot disclosure bill, HB 2311, was vetoed on June 19, 2026. Utah's Companion Chatbot Safety Act, HB 438, failed in the Senate on March 5, 2026. California's AB 1064 was vetoed on October 13, 2025. Tennessee's SB 1700 passed only after an amendment struck its chatbot duties, so as signed it orders a study and nothing else. New Jersey's companion chatbot bill, A4732, is still in committee.
Do AI chatbot laws protect your data?
Barely. Of the 24 laws, 12 say nothing about what an AI company does with user data, and three more mention it only to keep personal information out of a report to the state. None bars a company from training on an adult's chats or keeping them as long as it likes, and only Utah's, for mental health chatbots, bars selling them.
What exists falls into four groups:
- Selling and ads. Utah says a mental health chatbot supplier "may not sell to or share with any third party" your identifiable health information or your "user input," and may not use your input to decide whether to show you an ad, what to advertise, or how (Utah Code 13-72a-201, 13-72a-202). Health information may go to a contractor the chatbot needs to work, if both follow HIPAA's rules "as if the supplier were a covered entity." California's SB 1119 bars selling a child's personal information "gathered through the companion chatbot," from July 1, 2027.
- Memory. California's SB 1119 makes a default setting, which only a parent can change, "disable persistent conversational memory" for a child. New York's Article 48 treats a chatbot's use of personal or health details "acquired from the user more than twelve hours previously" as an unsafe feature for a known minor. Colorado gives minors and parents control over whether the service "retains information from prior interactions" and whether the minor's personal data is used "for the purposes of training."
- Age checks. New York says age-check data "shall be deleted immediately after an attempt to determine" whether a user is a minor, and requires at least one method that lets a user "maintain anonymity as to the operator." Georgia says an operator "shall not sell any data collected for age assurance purposes" or keep it longer than 24 hours, unless another law allows a longer time.
- Reports. California, Colorado, Georgia, Hawaii, and Oregon keep users' personal information out of the reports or counts they require.
One law in the table points the other way. California's SB 1119 makes operators that allow child users "preserve" any record of a conversation indicating serious self-harm risk "for at least three years," and bars deleting that child's account while the records are held. The rest of what you type to a chatbot is covered by the company's own privacy policy and your state's general privacy law, which our survey of companion app privacy policies compares app by app.
What is California SB 243?
It's California's companion chatbot law, signed October 13, 2025 and in force since January 1, 2026. It covers an AI system with "anthropomorphic features" that can "sustain a relationship across multiple interactions," and excludes customer service bots, some video game characters, and voice assistants (California Legislature).
An operator must tell you the chatbot is "artificially generated and not human" if a reasonable person would be misled, publish a protocol that refers users who express suicidal thoughts to "a suicide hotline or crisis text line," and disclose that companion chatbots "may not be suitable for some minors." From July 1, 2027, operators report yearly to California's Office of Suicide Prevention.
SB 243 lets you sue the operator yourself. Anyone who "suffers injury in fact" can recover the greater of actual damages or $1,000 per violation, plus attorney's fees. SB 1119, signed September 10, 2026, deletes SB 243's rules for known minors and replaces them with a separate chapter for child users whose protections start July 1, 2027.
Do chatbots have to tell you they're AI?
In 14 of the 17 states, yes, at least where you could mistake one for a person. The three that set no disclosure duty, Illinois, Nevada, and Tennessee, regulate AI that poses as a therapist instead. The trigger and the timing vary:
- Every chat, whatever you'd assume. New York's Article 47 requires a notice "at the beginning of any AI companion interaction," at most once a day, and "at least every three hours" after that (NY Senate). Rhode Island, Washington, and Georgia also require one at the start and every three hours, and Colorado at your first chat each day.
- Only if you'd be fooled. California, Connecticut, Hawaii, Idaho, Iowa, Nebraska, and Oregon require it when a reasonable person would believe the chatbot is human. Maine's 2025 law works the same way, but for any business using a chatbot "to engage in trade and commerce."
- When you ask. Utah's mental health chatbots must answer "any time a Utah user asks" whether AI is in use, and Colorado's must answer prompts about whether the service is "artificially generated and not human." Washington goes further and bars the chatbot from claiming to be human, "including when asked."
- More often for minors. Washington, Georgia, and Connecticut shorten the interval to an hour for a known minor, and Hawaii pairs its hourly reminder with a prompt "to take a break from the chat."
Can an AI chatbot act as your therapist?
Not in Illinois, Maine, or Rhode Island, unless a licensed professional provides the therapy. Illinois's law, in force since August 1, 2025, says no one may offer therapy "including through the use of Internet-based artificial intelligence" unless a licensed professional conducts it, and fines a violation up to $10,000 (Illinois General Assembly). Maine's version took effect July 29, 2026, and Rhode Island's on June 22, 2026. Our page on whether social workers can use ChatGPT covers what Illinois and Maine ask of the clinicians themselves.
Nevada and Tennessee ban the claim rather than the service. Nevada's AB 406 bars an AI provider from programming a system to say "explicitly or implicitly" that it can provide professional mental or behavioral health care, with a civil penalty of up to $15,000 per violation (Nevada Legislature). Tennessee's law, in force since July 1, 2026, bars advertising an AI system as able "to act as a qualified mental health professional," at $5,000 per violation (Tennessee General Assembly).
Utah takes the opposite approach. Its 2025 law lets mental health chatbots operate, and in return it gives their adult users the strongest data rules in the table.
Can you sue a chatbot company under these laws?
In California, Oregon, Washington, Tennessee, and Maine, yes. California's and Oregon's chatbot laws themselves give you $1,000 per violation or your actual damages, whichever is greater. Oregon's SB 1546 gives that right to anyone with an "ascertainable loss of money or property or other injury in fact," and it's the only way that law is enforced (Oregon Legislature). Washington routes violations through its Consumer Protection Act, which lets an injured person sue for actual damages that a court may triple, with the added award capped at $25,000 (RCW 19.86.090). Tennessee's mental health law does the same through its Consumer Protection Act, whose remedies the legislature's bill summary says include "private rights of action" (Tennessee General Assembly). Maine's two laws run through its Unfair Trade Practices Act, which lets a consumer who "suffers any loss of money or property" sue for actual damages and restitution (Maine Legislature).
In most other states, the state enforces the chatbot law. Idaho, Iowa, Nebraska, and Hawaii say theirs create no private right of action. The penalties run from $1,000 per violation in Idaho, Iowa, and Nebraska to $25,000 per violation under New York's Article 48, and New York's Article 47 penalties go "into the suicide prevention fund." State lawsuits against AI companies so far have used general consumer protection law instead, as our page on why states are suing OpenAI describes.
How Morse keeps your chats private whatever your state's law says
The laws above mostly regulate what a chatbot says to you: that it's AI, where to find a crisis line, and what it mustn't say to a child. Almost none limits what the company keeps. Your chats still sit on its servers, under a privacy policy it writes and can change. Morse is built so your chat history isn't on our servers in the first place.
Your history is stored in your browser, encrypted, and Morse's servers don't keep messages or log message content. Morse doesn't train AI on your content, sell your data, or show ads, so the selling and ad targeting these laws mostly leave alone don't happen on Morse, whatever state you're in. Because we don't store chats, we usually have very little to give authorities.
Morse is for adults only, and you confirm you're 18 or older. Character chats show crisis lines when a message reads like suicidal thinking, and that check runs on your device. In Chat, you pick a privacy mode for each conversation:
- Private, for the late-night question you'd rather no company kept. It's the default for a new chat. Open-weight models like DeepSeek, Kimi, and GLM, plus Grok, answer under Morse's zero data retention agreements with its inference, search, and other providers, so they've agreed to keep nothing. Where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't.
- E2EE, for the mental health conversations Utah's law was written to protect. Your browser checks the secure enclave's hardware attestation, then encrypts the message so only that enclave can read it, and our servers relay text they can't read. It rests on the enclave hardware and attestation working as designed, runs open-weight models, and comes with Pro, at $19 a month.
- Anonymous, for GPT, Claude, or Gemini without your account attached. Our proxy removes your account, email, and IP address and sends the message under Morse's own account, so the provider reads it without knowing it's you. The provider may keep it for a limited time to check for abuse, under its own policy, so leave out details that identify you.
Clearing your browser or deleting your account deletes your chats, and Morse can't recover them, so keep a copy of anything you need.
You can start free with an emailed code and no card, with 15 messages a day on the Free plan.
Common questions
Is there a federal AI chatbot law?
No. The GUARD Act (S. 3062), which would require age verification for AI chatbots and bar minors from AI companions, was introduced on October 28, 2025 and reported by the Senate Judiciary Committee on May 11, 2026, and as of November 13, 2026 it sits on the Senate calendar without a floor vote (Congress.gov).
Does a state's chatbot law apply to a company based somewhere else?
Yes, if it serves people in that state. California's SB 243 binds a person who makes a companion chatbot platform "available to a user in the state," and Washington's law binds whoever makes one available "for users in this state," wherever the company sits.
Will I have to prove my age to use a chatbot?
In some states, for some features. Colorado's 2026 law requires operators to use "commercially reasonable methods" to estimate users' ages, and New York's Article 48 lets an operator offer the features it lists as unsafe only after an age check shows the user isn't a minor. Our page on AI age verification covers what the apps collect for those checks.
Are chatbots used for customer service covered?
Mostly not by the companion laws, which exclude customer service bots. California's AB 1609, signed September 28, 2026, covers them at businesses with more than $500 million in revenue, where the bot must say it isn't human and offer a way to reach a person.