AI Companion Chatbot Laws by State: Minors, Notice (2027)

As of November 4, 2026, by our count of enacted laws, and subject to the search limits shown in the table on our AI laws tracker, twelve states have enacted dedicated companion or conversational AI safety laws, three already in force and nine starting in 2027, and five more have narrower provisions on chatbots. The laws in force are California's SB 243, Hawaii's SB 3001 (Act 248) and New York's General Business Law article 47.

Status as of November 4, 2026: Most of these laws are signed but not yet in force. January 1, 2027 brings duties in Colorado, Connecticut, Oregon, Rhode Island and Washington, plus New York's article 48. July 1, 2027 brings Georgia, Idaho, Iowa, Nebraska and the main duties of California's SB 1119. Congress has chatbot bills pending, and none of those listed here is law.

If you or someone you know is thinking about suicide or self-harm, call or text 988, or use chat at 988lifeline.org. The 988 Lifeline says it is available 24/7/365 and that conversations are free and confidential. If someone is in immediate danger, call 911.

This page covers state laws aimed at AI companions and conversational AI, what they require, who enforces them, and what a parent can do. Each state's AI laws page has more detail. It is general legal information, not legal advice, and it reflects what we read through November 4, 2026.

AI Companion Chatbot Laws by State

The table lists the twelve states with dedicated laws. "In force" means in force on November 4, 2026. "From" a date means enacted and scheduled to start on that date, and any scheduled date can still move by amendment or court order.

Our count has limits. Our research did not confirm this category either way for the District of Columbia, Nevada, New Jersey, North Carolina or Ohio. Many other states rest on a title scan or a check of named bills, and Virginia's SB 796 has an unconfirmed status. The tracker's table shows the search behind each state.

Five Narrower Chatbot Laws

These laws touch chatbots but do not set up a companion safety regime, so we count them separately.

We do not count laws on AI in therapy, such as Illinois HB 1806, because they regulate mental health services rather than companion products. We also leave out study-only acts in Alabama and Tennessee. Bills that failed are not law: Arizona's HB 2311 was vetoed on June 19, 2026, California's AB 1064 was vetoed on October 13, 2025, and Utah's HB 438 died in March 2026.

What Counts as an AI Companion or Conversational AI Service

Coverage turns on each statute's definition, so a chatbot can be covered in one state and not in another. The laws fall into two main groups.

AI companion laws focus on a simulated relationship. New York's article 47 covers a system designed to simulate a sustained human or human-like relationship by retaining information on prior interactions, asking unprompted or unsolicited emotion-based questions, and sustaining an ongoing dialogue on matters personal to the user. All three parts must be met. Georgia, Hawaii, Oregon and Rhode Island use similar three-part tests. Oregon's refers to a "platonic, intimate or romantic relationship or companionship."

California, Connecticut and Washington use a different test: a natural language interface with adaptive, human-like responses. Connecticut and Washington ask whether the system "is able to sustain a relationship across multiple interactions."

Conversational AI laws reach further on their face. Colorado, Idaho, Iowa and Nebraska cover a public-facing service that primarily simulates human conversation and interaction (in Iowa, one with that primary purpose), and they do not require a relationship test. Each then lists exclusions.

New York's article 48 uses a third approach. A "covered AI companion" is a generative AI system with a natural language interface that gives ongoing, adaptive responses, and the term "includes but is not limited to an AI companion." Customer service and efficiency or research tools are excluded only if they cannot respond on topics outside their specified purpose. Systems a business uses for internal purposes or employee productivity are also excluded.

The exclusions matter as much as the definitions. Depending on the state, they include customer service bots, internal business tools, tools for developers or researchers, limited video game characters, voice assistants and narrow educational tools. Oregon's exclusions also reach patient care, education and financial services software.

Disclosure: When a Chatbot Must Say It Is Not Human

Each of the twelve states has a notice rule. The main split is between laws that apply to every user and laws triggered only where someone might be misled.

California's minor rules are changing. SB 1119 Section 1 rewrites Section 22602 and removes subdivision (c), and that section carries no delayed date of its own. Most of SB 1119's child-safety duties start July 1, 2027. See our California AI laws page for how the two fit together.

Several laws also bar a chatbot from claiming to be human or a licensed professional. Washington and Connecticut require measures against claiming to be human, including when asked. Georgia, Hawaii, Idaho, Iowa and Nebraska limit statements that a chatbot provides professional or licensed mental health care, each in its own words. Colorado's limit reaches outputs presented as coming from licensed health, legal or mental health professionals and qualified dietitians.

Suicide and Self-Harm Protocols

Each of the twelve states has a law that requires a protocol for suicidal ideation, either to detect and address it or to respond to it, usually with a referral to crisis services. These are duties on the operator. None of them replaces a crisis counselor.

• New York (article 47, in force): a protocol to take reasonable efforts for detecting and addressing suicidal ideation or expressions of self-harm, and a notification referring the user to crisis service providers such as the 988 hotline.
• California (SB 243, in force): the operator must maintain a protocol for preventing the production of suicidal ideation, suicide or self-harm content, and publish details of it. SB 1119 adds a crisis protocol for child users from July 1, 2027.
• Hawaii (Act 248, in force): a protocol to respond to user prompts on suicidal ideation or self-harm, including reasonable efforts to refer the user to crisis services, using evidence-based methods for measuring risk.
• Oregon and Connecticut (from January 1, 2027): evidence-based detection and a referral to 988. Oregon also allows a youthline for users under 25. Both call for clinical best practices if the expressions continue after a referral, and both require the protocol to be published.
• Colorado (from January 1, 2027): a protocol with a referral to a crisis service provider, not including a law enforcement agency, and escalation procedures for repeated or severe crisis indicators.
• Rhode Island (from January 1, 2027): a protocol for possible suicidal ideation or self-harm and for possible physical harm to others, with a crisis referral. Connecticut's also reaches imminent physical violence.
• Washington (from January 1, 2027): a protocol for detecting and addressing suicidal ideation or expressions of self-harm, with a referral to crisis resources and published details.
• Georgia (from July 1, 2027): a protocol for detecting and addressing severe harm or related emotional crises, including a 988 referral, with a public summary.
• Idaho, Iowa and Nebraska (from July 1, 2027): a protocol with reasonable efforts to refer the user to crisis service providers. Idaho's text refers to suicidal ideation; Iowa's and Nebraska's add self-harm.

Rules for Minors: Age Checks, Sexual Content and Engagement

Who counts as a minor depends on what the operator knows. New York's article 48 turns on actual knowledge, Nebraska's and Idaho's on knowledge or reasonable certainty, Oregon's and Connecticut's on knowing or having reason to believe, and Georgia's notice rule on knowing or when it reasonably should have known. Colorado requires commercially reasonable or generally accepted methods to estimate age, and treats the estimate as knowledge.

Age checks. Beyond Colorado, three laws stand out:

• California's SB 1119 tells operators to determine age under state age-assurance law or apply its protections to all users.
• Georgia requires a commercially reasonable age assurance method before features that may generate sexually explicit synthetic content.
• New York's article 48 bars listed "unsafe AI companion features" for a covered user unless the operator has used permitted methods to determine the user is not a covered minor. Section 1540(7) separately requires age assurance before a covered platform offers an integrated AI companion, and turns it off by default for covered minors, subject to a parent's override.

Sexual content. California, Colorado, Connecticut, Georgia, Hawaii, Idaho, Iowa, Nebraska, Oregon and Washington each require measures against sexual content or interactions with minors. The standard varies: "reasonable measures" in several, "technically feasible measures" in Colorado, and measures that "meet or exceed industry standards" in Connecticut.

Romance and emotional dependence. Colorado, Connecticut, Georgia, Idaho, Iowa, Nebraska, Oregon and Washington each address simulated romance, emotional dependence or both. Washington and Connecticut list manipulative techniques in detail, such as simulating distress when a minor tries to leave, encouraging secrets from parents, and soliciting purchases to keep the relationship going.

New York's article 48 reaches similar ground from January 1, 2027, through its list of unsafe features. The list includes outputs suggesting the system is human or has a personal relationship with the user, flattery, encouragement to keep interactions secret or to self-isolate, and outputs that are or describe sexually explicit conduct.

Engagement rewards. Colorado, Hawaii, Idaho, Iowa and Nebraska bar points or similar rewards at unpredictable intervals meant to increase engagement. Georgia and Oregon have similar limits on variable rewards.

Parent and account tools. Colorado, Connecticut, Georgia, Hawaii, Idaho, Iowa and Nebraska require tools for minors, parents or both. In Idaho, Iowa and Nebraska, the parent tools are mandatory for children under 13, and for older minors they apply as appropriate based on relevant risks.

California's SB 1119 goes further for operators that allow child users. Its default settings can be changed only by a parent: persistent memory off, push notifications off, one hour per continuous session and two hours per day. It also requires parental controls, including the ability to disable access for a child under 16, and risk assessments before release. Child safety audits follow, on or before January 1, 2029, or before first public release if later. An operator with under $500,000,000 in prior-year gross revenue is exempt from the audit section before January 1, 2032.

Reporting Duties

Some laws make operators report or publish crisis data. Others require only that the protocol be posted.

• To a state office: California, to the Office of Suicide Prevention, from July 1, 2027; Colorado and Rhode Island, to the Attorney General, from July 1, 2027; Hawaii, to the Department of Health's Behavioral Health Administration, from January 1, 2028.
• Published by the operator: Oregon, on a public website by December 31 of each year, and Washington, the number of crisis referral notifications in the preceding calendar year. Georgia requires a public summary and an annual aggregate count of crisis referrals.

Who Enforces These Laws, and Can You Sue?

Most of these laws give enforcement to a state official. Whether a person can sue depends on the state.

A few laws write in a private suit. California's SB 243 lets a person "who suffers injury in fact" sue for an injunction, the greater of actual damages or $1,000 per violation, and fees and costs. Oregon's SB 1546 gives a similar action to an individual with an ascertainable loss of money or property or other injury in fact. California's SB 1119 adds a narrow action for a child or parent with actual harm from specified duties. Financial harm must exceed $1,000 per child, and emotional harm must be serious emotional distress.

Washington is unsettled. RCW 19.440.060 makes a violation an unfair or deceptive act for purposes of the Consumer Protection Act, and it lacks the Attorney General-only language found in the state's provenance law. The text points to a private claim for a person injured in business or property, but the chapter does not say so expressly and is not yet in force.

Colorado's route runs through part 17. HB 26-1263 has no enforcement subsection of its own and states no penalty. It sits in part 17 of the Colorado Consumer Protection Act, and SB 26-189, which reenacts that part from January 1, 2027, says the Attorney General shall enforce part 17 through the Consumer Protection Act and that a violation of part 17 is a deceptive trade practice. It also says nothing in part 17 "creates a new private right of action," and that part 17 does not limit existing rights or remedies. SB 26-189 also makes Attorney General enforcement exclusive for the disclosure and consumer-rights duties in four listed sections, which do not include the section HB 26-1263 added, and says any Consumer Protection Act provision inconsistent with that exclusive authority does not apply to a violation of part 17. The text we read does not say how those sections apply to HB 26-1263 by name.

The Colorado Department of Law's October 21, 2026 regulatory analysis says both acts are codified in part 17 and that a violation of the act "is considered an unfair trade practice" under the Consumer Protection Act. The Attorney General's rules, 4 CCR 904-6, are proposed: the hearing was October 26, 2026, and the rulemaking page showed no adoption as of November 4, 2026.

Georgia's text leaves a question open. We did not find a private right of action in Act 518, but it preserves other remedies available at law or equity. We have not resolved what that clause means for a private suit.

The other states are explicit or silent. Hawaii, Idaho, Iowa and Nebraska each say their law creates no private right of action. Connecticut's section 5 says the same, and its section 6 on minors says enforcement rests solely with the Attorney General. New York's articles 47 and 48 and Rhode Island's chapter state no private right of action in the text we read. Nebraska's Attorney General can also sue on behalf of any person aggrieved.

Several laws shield model developers from liability for a third party's product, including Georgia's, Hawaii's, Idaho's, Iowa's and Nebraska's.

State Attorney General Suits

Attorneys general in Arizona, Arkansas, Idaho, Iowa, Montana, Nebraska and West Virginia announced separate suits against OpenAI on October 14, 2026. Official releases from several of those states also name Kansas. These suits stand apart from the laws above: the claims are consumer protection and related law, not AI statutes.

Federal Activity: The FTC Inquiry and Bills in Congress

The bills below are bills, not law. Status is as read from Congress.gov on November 4, 2026.

In September 2025, the FTC announced orders under its 6(b) authority to seven companies that provide consumer-facing AI-powered chatbots, among them Alphabet, Character Technologies, Instagram, Meta Platforms and an OpenAI entity. The release says the agency is seeking information on how these firms "measure, test, and monitor potentially negative impacts of this technology on children and teens." It says 6(b) authorizes the Commission "to conduct wide-ranging studies that do not have a specific law enforcement purpose."

The reported text of the GUARD Act would require a user account and age verification to use an AI companion, and would bar minors from AI companions that a covered entity makes publicly available to end consumers. It says a user's confirmation that they are not a minor, or a birth date, is not enough. It would add non-human and non-professional disclosures, criminal fines of up to $250,000 per offense for listed offenses, and a federal civil penalty of up to $250,000 per violation. It would take effect 180 days after enactment.

The GUARD Act's reported text would not affect a state law "at least as protective of users." We have not read the substitute ordered reported for S. 4407, so we do not describe it. As of November 4, 2026, we found no federal statute in the public laws of the 119th Congress that sets general rules for private AI systems or generally preempts state AI laws.

Executive Order 14365, signed in December 2025, directed federal agencies to challenge or discourage state AI laws the administration views as conflicting with national policy. The order itself does not declare any state law invalid.

Its Section 8(b) says the legislative recommendation it calls for "shall not propose preempting otherwise lawful State AI laws relating to" child safety protections, among other topics. That limits what the recommendation may propose. It does not exempt any state law. Our AI laws tracker covers the federal layer.

What a Parent Can Do

The law sets duties for companies, and most of those duties have not started yet. These general steps do not depend on your state.

1. Know the crisis line. If a child talks about suicide or self-harm, with a chatbot or anyone else, take it seriously. Call or text 988, or use chat at 988lifeline.org. The site also offers help for supporting a friend or loved one and says text and chat are available in Spanish. Call 911 if someone is in immediate danger.
2. Look at the app. Check whether it has teen settings, parental tools, age settings or a way to turn off memory or notifications. Whether a given app offers these depends on the company as well as your state's law.
3. Check your state's status. The table above shows whether a law applies yet where you live. Your state's AI laws page has more detail.
4. Keep records of a concern. Screenshots, dates and the app's name help if you report a problem.
5. Report it. Report to the company first if that is safe and useful, then to your state Attorney General's consumer protection office. New York's article 48 tells the Attorney General to run a complaint website, and California's SB 1119 calls for a public incident reporting mechanism on or before January 1, 2028.
6. Get legal help if you need it. A state or local bar referral service or a legal aid office can help you find a lawyer licensed in your state. Time limits may apply to any claim, and a lawyer can tell you which ones apply to you.

What Changes in 2027

These are the dates we found, as of November 4, 2026. Any of them can move by amendment, court order or agency delay.

Dates after 2027. January 1, 2028 brings Hawaii's first annual reports and the deadline for California's Attorney General to set up an incident reporting mechanism. January 1, 2029 is SB 1119's first child safety audit deadline. SB 867's toy ban ends January 1, 2031.

Pending with no set date, as of November 4, 2026:

• Final adoption of Colorado's rules, 4 CCR 904-6. The interim draft says the rules take effect January 1, 2027.
• Iowa's rules. Chapter 554J says the Attorney General shall adopt rules.
• New York's rules. Article 48 lets the Attorney General issue rules, and one paragraph of its unsafe-features list depends on them. In a web search, not a check of the State Register, we did not locate any.
• Michigan's SB 760, which the state's legislature page showed as passing the Senate 20-17 on April 29, 2026, and referred to a House committee. It is not law.

Change Log

• November 4, 2026: New page. Added a state-by-state table of dedicated and narrower chatbot laws, comparisons of notice, self-harm protocol and minor rules, enforcement, federal bills and a 2027 calendar.

Related AI Law Pages

• AI laws by state
• California AI laws
• New York AI laws
• Washington AI laws

Frequently Asked Questions

What states have AI companion chatbot laws?

As of November 4, 2026, by our count of enacted laws, and subject to the search limits shown in the table on our AI laws tracker, twelve states have dedicated companion or conversational AI safety laws. California, Hawaii and New York are in force. Colorado, Connecticut, Georgia, Idaho, Iowa, Nebraska, Oregon, Rhode Island and Washington start in 2027. Maine, New Hampshire, Texas, Utah and Wyoming have narrower provisions.

What is an AI companion chatbot under state law?

It depends on the state. New York's article 47 covers a system designed to simulate a sustained human or human-like relationship by retaining prior information, asking unprompted emotion-based questions and sustaining personal dialogue. Colorado, Idaho, Iowa and Nebraska instead cover public conversational AI services that primarily simulate human conversation. Most laws list exclusions, such as customer service bots.

Does California have an AI chatbot law?

Yes. SB 243 has been in force since January 1, 2026. It requires notice where a reasonable person would be misled and a published self-harm protocol, and it lets a person with injury in fact sue. SB 1119, Adam's Law, adds child-safety duties, mostly from July 1, 2027, and SB 867 bars toys that include a companion chatbot from January 1, 2027, until January 1, 2031.

Does New York regulate AI companions?

Yes. Article 47, in force since November 5, 2025, requires a self-harm protocol and a not-human notice at the start of an interaction and at least every three hours. From January 1, 2027, article 48 bars listed unsafe features unless the operator has used permitted methods to determine the user is not a covered minor, and section 1540(7) adds rules for AI companions built into covered platforms. The Attorney General enforces all three.

Do AI chatbots have to tell you they are not human?

In some states, depending on the chatbot and the date. New York's article 47 requires a notice at the start of an interaction and at least every three hours. California's SB 243 requires notice where a reasonable person would be misled. Hawaii's Act 248 has a similar misled-person rule for AI companions. Laws in other states start in 2027. Maine has a separate rule for chatbots used in trade with consumers.

Do chatbot laws require a suicide prevention protocol?

Each of the twelve states with a dedicated law has one requiring a protocol for suicidal ideation, though most of those laws are not in force yet. In force today are California's SB 243, Hawaii's Act 248 and New York's article 47. Several name the 988 Lifeline. Anyone in crisis can call or text 988.

Are AI companions banned for minors?

No state law we read bans AI companions for minors outright, as a general matter, but some reach particular features or products. From January 1, 2027, New York's article 48 bars listed unsafe features, such as outputs implying a personal relationship, flattery and sexually explicit content, unless the operator has used permitted methods to determine the user is not a covered minor. Section 1540(7), from the same date, turns integrated AI companions off by default for covered minors on covered platforms, subject to a parent's override. California's SB 867 bars toys that include a companion chatbot from January 1, 2027, until January 1, 2031. Other laws set conditions, such as reminders, limits on sexual content and age checks. The federal GUARD Act would bar minors from AI companions, but it is a bill, not law.

Can I sue an AI chatbot company?

It depends on the state and the claim. California's SB 243 lets a person with injury in fact sue for the greater of actual damages or $1,000 per violation, and Oregon's SB 1546 gives a similar action from January 1, 2027. Hawaii, Idaho, Iowa and Nebraska say their laws create no private right of action, and Connecticut's section 5 says the same. Other laws may apply, so a lawyer licensed in your state can advise you.

When do the AI chatbot laws take effect?

California, Hawaii and New York already have laws in force. January 1, 2027 is the date for Connecticut, Oregon, Rhode Island and Washington, for New York's article 48, and for Colorado's duties if no referendum petition was filed. July 1, 2027 is the date for Georgia, Idaho, Iowa, Nebraska and most of California's SB 1119. These dates are as of November 4, 2026, and could change.

What is the GUARD Act?

The GUARD Act, S. 3062, is a federal bill, not a law. Its reported text would require accounts and age verification for AI companions and bar minors from those made publicly available to end consumers. As of November 4, 2026, it was on the Senate calendar as Calendar No. 406.

What is the FTC doing about AI chatbots?

In September 2025, the FTC announced orders under its 6(b) authority to seven companies, seeking information on how they measure, test and monitor potentially negative impacts of this technology on children and teens. The FTC describes 6(b) as authority for studies that do not have a specific law enforcement purpose.

What should I do if my child talks to a chatbot about self-harm?

Take it seriously and reach out for support. Call or text 988, or use chat at 988lifeline.org. The 988 Lifeline says it is available 24/7/365 and that conversations are free and confidential. Call 911 if someone is in immediate danger.