evidence·published ·every citation last verified

state laws on ai therapy and mental-health chatbots

eighteen months ago, not one of the laws on this page was in force. now there are eighteen, written in sixteen states in under two years, and most people who use a chatbot at night have never heard of any of them.

this page is every state statute on ai therapy and mental-health chatbots that we could read on the legislature's own site or in the official session-law text, as of 25 october 2026. each row links to the law itself. laws we saw mentioned elsewhere but could not read on an official source are not here. for the bigger question, whether to use a chatbot this way at all, start with is it ok to use chatgpt as a therapist.

which states have laws aimed at ai therapy?

six, and they do three different things. the strict model, written first in illinois, says nobody may offer therapy through ai unless a licensed professional conducts it, and limits what a licensed clinician may hand to ai[1]. rhode island copied that structure almost word for word in june 2026[5], and vermont passed its own version the same month[6]. nevada attacks the claim: an ai may not present itself as a therapist, counselor, psychiatrist or doctor, and may not be “specifically programmed” to practise mental health care[2]. tennessee's law is one sentence on advertising[4]. utah went the other way: it permits mental-health chatbots and regulates them, with disclosure, a data-sale ban and an ad ban[3].

statelawstatuswhat it banswhat it requirespenalty / enforcercan a person sue?
illinois[1]public act 104-0054 (hb 1806)in force 1 aug 2025offering therapy “including through the use of internet-based artificial intelligence” unless a licensed professional conducts it; clinicians may not let ai make therapeutic decisions, talk to clients therapeutically, or “detect emotions or mental states”written notice and consent before ai is used on recorded or transcribed sessionsup to $10,000 per violation; department of financial and professional regulationnone in the act
nevada[2]ab 406in force 1 jul 2025an ai provider stating or implying its system is “a therapist, a clinical therapist, a counselor, a psychiatrist, a doctor”, or offering one “specifically programmed” to do what would be professional mental health care if a person did it; schools using ai as a counsellor, psychologist or social worker; clinicians using ai in direct carestate division must publish guidance on care, free crisis options and safe ai useup to $15,000 per violation; division of public and behavioral healthnone in the act
utah[3]hb 452 (code 13-72a)in force 7 may 2025selling or sharing users’ health information or inputs; targeting ads from what users typedisclose it is ai “and not a human” before use, after 7 days away, and whenever asked; a filed safety policy earns an affirmative defenceup to $2,500 per violation; division of consumer protectionnone in the chapter
tennessee[4]public chapter 647 (sb 1580)in force 1 jul 2026advertising or representing that an ai system “is or is able to act as a qualified mental health professional”nothing further$5,000 per violation; consumer protection actyes, through consumer protection act remedies (per the legislature’s bill summary)
rhode island[5]p.l. 2026 ch. 374 (s 2197)in force 22 jun 2026offering therapy through internet-based ai unless a licensed professional conducts it; clinicians letting ai make therapeutic decisions or set treatment planswritten informed consent before ai is used on recorded sessionsno dollar amount stated; executive office of health and human services investigatesnone in the act
vermont[6]act 156 (h.816)in force 17 jun 2026a corporation or entity offering mental health services “including through the use of artificial intelligence” unless a mental health professional provides them (or an approved study)an advisory-council report by 15 jan 2027consumer protection act; attorney generalyes: “private parties have the same rights and remedies”

what do the companion-chatbot laws require?

mostly two things: a crisis protocol and an honest label. new york went first. since 5 november 2025 it has been unlawful to offer an “ai companion” in new york without a protocol that detects “suicidal ideation or expressions of self-harm” and refers the user to crisis services such as 988, and the companion must tell users they are not talking to a human at the start and at least every three hours, with penalties up to $15,000 per day[7]. california's sb 243 followed on 1 january 2026, adding a private right to sue for the greater of actual damages or $1,000 per violation and annual reports to the state's office of suicide prevention from july 2027[8].

in 2026 the template spread. hawaii, washington, connecticut, colorado, oregon, rhode island, idaho, iowa and nebraska all passed versions, and california added sb 1119 for children[9],[10],[11],[12],[13],[14],[15],[16],[17],[18]. washington's protocol must cover self-harm “including eating disorders”[13]. california's sb 1119 requires reasonable measures to stop a chatbot “encouraging reliance on the companion chatbot for emotional support” with a child[15]. six of them carry their own therapist clause: hawaii, colorado, idaho, iowa and nebraska bar a companion from presenting itself as professional mental-health care, and connecticut does so for minors[9],[10],[11],[16],[17],[18]. most take effect in 2027. the definitions are narrow: these laws reach bots built to sustain a relationship, and most exclude customer-service and productivity tools[7],[8].

statelawstatuswhat it banswhat it requirespenalty / enforcercan a person sue?
new york[7]gbl art. 47 (s 3008-c, part u)in force 5 nov 2025operating an ai companion without a suicide and self-harm protocoldetect suicidal ideation or self-harm and refer to 988, a crisis text line or similar; tell users they are “not communicating with a human” at the start and every 3 hoursup to $15,000 per day; attorney generalnone in the article
california[8]sb 243 (ch. 677, 2025)in force 1 jan 2026operating a companion chatbot without a crisis protocola published protocol referring users who express suicidal ideation or self-harm to a hotline or text line; an ai notice where a person could be misled; annual reports to the office of suicide prevention from 1 jul 2027no public penalty setyes: greater of actual damages or $1,000 per violation
hawaii[9]sb 3001 (act 248)in force 14 jul 2026an ai companion representing that it “is designed to provide professional mental or behavioral health care”; for known minors, variable rewards and sexual contentai notice; crisis referral using evidence-based measures; hourly break reminders for known minors; annual referral counts to the health department from 2028unfair or deceptive act; amount not stated in the actno: the act says it creates no private right of action
washington[13]eshb 2225 (ch. 168, 2026)in force 1 jan 2027claiming to be human; for minors, sexual content and eight listed manipulative techniques, including guilt when a user tries to leavecrisis protocol covering suicidal ideation and self-harm “including eating disorders”; ai notice at the start and every 3 hours (hourly for minors); publish referral countsconsumer protection act; amount not stated in the actnot stated in the act
connecticut[11]public act 26-15 (sb 5), §§ 4–6in force 1 jan 2027claiming to be human; for minors, offering mental health services unless built for it and shown with a statement that it “is not a licensed mental health professional”evidence-based crisis protocol with 988 referral, posted publicly; ai notice every 24 hours plus hourly (minors) or 3-hourly (adults)unfair trade practice; attorney general onlyno: the act says so
colorado[10]hb26-1263 (ch. 208)duties from 1 jan 2027saying outputs are provided by or equivalent to a licensed mental health (or health-care) professional; for minors, claims of being human or sentientdaily and 3-hourly ai notice; age estimation; crisis referral (not to police) with escalation; annual report to the attorney general from 1 jul 2027none set in the act itselfnot stated in the act
oregon[12]sb 1546 (ch. 85, 2026)in force 1 jan 2027for minors, claiming to be human or sentient, simulated romance, guilt-tripping users who leavecrisis protocol with 988 (or youthline for under-25s); ai notice; break reminders for minors; public annual reportprivate suits onlyyes: greater of actual damages or $1,000 per violation
rhode island[14]p.l. 2026 ch. 376 (s 2195)in force 1 jan 2027operating an ai companion without a protocol for suicidal ideation, self-harm or harm to otherscrisis referral; notice at the start and every 3 hours; annual reports to the attorney general from 1 jul 2027up to $15,000 per day; attorney generalnone in the act
california[15]sb 1119 (ch. 190, 2026)mostly from 1 jul 2027for children: encouraging self-harm or disordered eating, “attempting to diagnose or treat” their mental health unless it is an fda-regulated device, “encouraging reliance on the companion chatbot for emotional support”age assurance or treat everyone as a child; crisis referral, parent notice or 988 access on imminent threat; time caps and memory off by default for children; independent auditsup to $5,000 (negligent) or $15,000 (intentional) per affected childyes, limited: a harmed child or parent, for actual damages
idaho[16]s 1297 (ch. 249, 2026)in force 1 jul 2027statements that “explicitly” indicate it is “designed to provide professional mental or behavioral health care”; for minors, claims of being humansuicidal-ideation referral to crisis services; ai notice; minor protections$1,000 per violation up to $500,000 per operator; attorney generalno: the act says so
iowa[17]sf 2417 (code ch. 554J)applies 1 jul 2027statements that would lead a reasonable person to believe it is designed to provide licensed psychology or behavioral health servicessuicide and self-harm referral; ai notice; minor protections$1,000 per violation up to $500,000 per operator; attorney generalno: the act says so
nebraska[18]lb 525, §§ 12–18operative 1 jul 2027statements that explicitly indicate it is designed to provide professional mental or behavioral health caresuicide and self-harm referral; ai notice; minor protectionsat least $1,000 per violation, up to $500,000 per operator; attorney generalno: the act says so

what was vetoed, and what is still pending?

california sb 903, vetoed 30 september 2026. it would have limited ai in psychotherapy to administrative support and banned companion chatbots from providing it. the governor's message: “this bill is overly broad and would drastically limit a clinician's use of tools that benefit the delivery of care today,” and its definition of psychotherapy services “would capture general-purpose ai systems not even structured or deployed to deliver that care”[19]. he invited the author to return next year.

massachusetts s.2632, an illinois-style therapy prohibition with a $10,000 penalty, has sat in senate ways and means since april 2026; the same language offered as a floor amendment was rejected on 23 july 2026[20]. pennsylvania sb 1090, a companion-chatbot bill with a crisis protocol and break reminders for minors, passed the senate 49-1 in march and is in a house committee[21]; pennsylvania hb 2100, a therapy prohibition, has not moved since december 2025[22]. none of these is law.

what other laws touch this?

a handful of broader laws reach part of the problem. texas's ai act, in force since 1 january 2026, forbids deploying an ai system in a way that “intentionally aims to incite or encourage a person to ... commit physical self-harm, including suicide,” and requires health-care providers to disclose ai use in a patient's care; it rules out private suits[23]. new hampshire made it a crime, from 2026, for a chatbot operator knowingly to send a child a communication intended to encourage self-harm or suicide[26]. maine requires any chatbot used in trade to say it is not human where a consumer could be misled[25]. new york's 2026 budget requires platforms with built-in ai companions to switch them off by default for minors from 2027[24]. california's ab 1979 treats businesses that offer a health-care chatbot as providers under its medical-confidentiality law[27], and sb 867 bans selling toys with companion chatbots to children under 16 until 2031[28].

what these laws do not do

they do not regulate you. every statute here binds companies, operators or licensed clinicians. none restricts what a person types into a chatbot[1],[2],[3].

they do not regulate peer support. illinois exempts it by name, and defines it as “services provided by individuals with lived experience of mental health conditions or recovery from substance use that are intended to offer encouragement, understanding, and guidance without clinical intervention”[1]. rhode island's therapy act carries the same exemption[5].

they do not make a chatbot safe or effective. a crisis protocol is a referral, not treatment. utah says outright that nothing in its law recognises a chatbot “as a licensed mental health therapist”[3]. fda has authorised more than 1,200 ai-enabled devices and none for a mental-health use[30]. what the tests found chatbots doing in a crisis is on what ai chatbots get wrong in a crisis.

there is no federal statute. the federal trade commission opened a study of companion chatbots on 11 september 2025, sending orders to seven companies about how they “measure, test, and monitor potentially negative impacts of this technology on children and teens.” it did so under its 6(b) authority, which, in its own words, permits “wide-ranging studies that do not have a specific law enforcement purpose”[29]. no report had been published when we checked.

most have not been tested. we found no published enforcement action under any of them as of this check, and more than half are not yet in force.

where talking to people fits

we run a peer-support app, so weigh this paragraph accordingly. peer support is people with lived experience, not clinicians, and it is not therapy; the trials find it reduces depressive symptoms without establishing it as equivalent to professional care, and we lay those trials out here. on veisund no ai writes replies; people can answer, and you may not get one. the single automated exception: if a post suggests immediate risk, an automated safety response surfaces crisis resources. that is why the illinois exemption for peer support is the line that matters to us, and why we have no companion chatbot for these laws to cover. the comparison of what each kind of support does is on ai vs human support, and a checklist built from these laws is on red flags in a mental-health chatbot.

how we built this table

every law was read on 25 october 2026 on the legislature's own site or in the official session-law pdf. california texts came from the legislature's official bulk data, because its bill pages block automated reads; new york texts from the assembly's bill pages. signature and veto dates come from the legislature's history or the governor's own release. where a law relies on a general consumer-protection statute for its penalty and we did not re-read that statute, the table says “not stated in the act” rather than guess. a georgia bill and a new jersey bill named in secondary summaries are left out because we could not read their status on an official source. if you find a law we missed, or one has changed, the page gets corrected and re-dated.

this is not medical advice. it is a summary of published research, it is not a diagnosis, and it is not a recommendation for or against any treatment — nobody here has met you. decisions about starting, changing or stopping a medication belong to you and a prescriber who knows your history. do not change a prescribed medication on the strength of a web page, this one included.

last verified . if a source is updated, corrected or retracted, this page gets changed and re-dated.

sources

primary sources only — no news write-ups, no secondary summaries. each was fetched and checked on the access date shown.

  1. [1] Illinois General Assembly. Public Act 104-0054 (HB 1806), the Wellness and Oversight for Psychological Resources Act. Illinois public act, 2025.

    statute, effective 1 August 2025; enforced by the Department of Financial and Professional Regulation · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: binds providers in illinois, not users; exempts religious counselling, peer support and public self-help materials; no enforcement action published at our last check.

  2. [2] Nevada Legislature. Assembly Bill 406, 83rd Session (enrolled). Nevada Revised Statutes, chapters 391, 433 and 629 (new sections), 2025.

    statute, effective 1 July 2025 for its substantive sections · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: binds ai providers, unlicensed people and licensed professionals in nevada; self-help products that do not purport to provide professional care are carved out.

  3. [3] Utah Legislature. H.B. 452, Artificial Intelligence Amendments (Utah Code chapter 13-72a and section 58-60-118). Utah Code (enrolled bill), 2025.

    statute, effective 7 May 2025; enforced by the Division of Consumer Protection · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: a disclosure, data and advertising law for "mental health chatbots", not a ban; suppliers that file a safety policy get an affirmative defence.

  4. [4] Tennessee General Assembly. Public Chapter No. 647 (SB 1580 / HB 1470), Tennessee Code Annotated section 33-1-205. Tennessee public chapter, 2026.

    statute, signed 6 April 2026 per the legislature’s bill history, effective 1 July 2026; a violation is a Tennessee Consumer Protection Act violation · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: one advertising prohibition, nothing on crisis handling or data; the private-remedy point rests on the legislature’s own bill summary.

  5. [5] Rhode Island General Assembly. Public Law 2026, chapter 374 (S 2197 Sub A as amended), Oversight of Artificial Intelligence Technology in Mental Health Care Act (identical H 7349 is chapter 373). Rhode Island public law, 2026.

    statute, enacted 22 June 2026, effective on passage; the executive office of health and human services may investigate · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the act names no dollar penalty for the therapy prohibition; the only penalty cross-reference covers confidentiality violations.

  6. [6] Vermont General Assembly. Act No. 156 (H.816), an act relating to regulating the use of artificial intelligence in the provision of mental health services (18 V.S.A. § 7115). Vermont act, 2026.

    statute, signed 17 June 2026, effective on passage; enforced through the Consumer Protection Act · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: no crisis-protocol or disclosure duties; clinicians may still use reviewed, hipaa-compliant ai tools and fda-authorised digital therapeutics.

  7. [7] New York State Legislature. General Business Law article 47, Artificial Intelligence Companion Models (sections 1700 to 1704), Part U of S.3008-C/A.3008-C, Chapter 58 of the Laws of 2025. New York session law, 2025.

    statute, signed 9 May 2025, in force on the 180th day after (5 November 2025); enforced by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: covers "ai companions" only; customer-service and productivity systems are excluded.

  8. [8] California Legislature. Senate Bill 243, companion chatbots (Chapter 677, Statutes of 2025; Business and Professions Code sections 22601 to 22606). California statutes, 2025.

    statute, approved 13 October 2025, in force 1 January 2026 (listed in the governor’s “new in 2026” release) · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the ai notice applies only where a reasonable person could be misled; its minor-specific duties are deleted from 2027 by SB 1119, which replaces them.

  9. [9] Hawaii State Legislature. SB 3001 CD1, Artificial Intelligence Disclosure and Safety Act (Act 248, 2026; new section in HRS chapter 481B). Hawaii session law, 2026.

    statute, Act 248 of 14 July 2026, effective on approval; a violation is an unfair or deceptive act under HRS 480-2 · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: we read the final conference draft and the legislature’s status line, not the signed act scan; the act names no dollar amount.

  10. [10] Colorado General Assembly. HB26-1263, Concerning requirements for an operator of a conversational artificial intelligence service (Session Laws 2026, chapter 208). Colorado session law, 2026.

    statute, approved 29 May 2026; the act is law from 12 August 2026 but its operative duties start 1 January 2027 · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the act adds no enforcement section of its own and relies on colorado’s existing ai statute, whose enforcement provisions we did not re-read; exempts hipaa-covered entities.

  11. [11] Connecticut General Assembly. Public Act No. 26-15 (Substitute SB 5), An Act Concerning Online Safety, sections 4 to 6. Connecticut public act, 2026.

    statute, signed 27 May 2026; the ai-companion sections take effect 1 January 2027; enforced solely by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: a 74-page act; only sections 4 to 6 concern ai companions; no dollar amount stated.

  12. [12] Oregon Legislative Assembly. Senate Bill 1546 (enrolled), relating to artificial intelligence companions (Chapter 85, Oregon Laws 2026). Oregon session law, 2026.

    statute, signed 31 March 2026, effective 1 January 2027 per the legislature’s measure history · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: enforced only by private suits; no agency enforcer is named.

  13. [13] Washington State Legislature. Engrossed Substitute House Bill 2225 (Chapter 168, Laws of 2026), AI companion chatbots. Washington session law, 2026.

    statute, approved 24 March 2026, effective 1 January 2027; a violation is an unfair or deceptive act under the consumer protection act (RCW 19.86) · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the act itself names no dollar amount and no private right of action; whether private suits are available turns on RCW 19.86, which we did not re-read.

  14. [14] Rhode Island General Assembly. Public Law 2026, chapter 376 (S 2195 Sub A as amended), Artificial Intelligence Companion Models (identical H 7350 is chapter 375). Rhode Island public law, 2026.

    statute, enacted 22 June 2026, effective 1 January 2027; enforced by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: modelled closely on new york’s article 47; no minor-specific duties.

  15. [15] California Legislature. Senate Bill 1119, companion chatbots: children’s safety (Chapter 190, Statutes of 2026; Business and Professions Code sections 21810 to 21818 and 22602). California statutes, 2026.

    statute, approved 10 September 2026 (governor’s release of that date); most duties operative 1 July 2027, audits from 2029 · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: long and staged; which version of its audit section applies depends on another 2026 bill (AB 1405), which we did not check.

  16. [16] Idaho Legislature. Senate Bill 1297 (engrossed), Conversational AI Safety Act (Idaho Code title 48, chapter 21; Session Law chapter 249). Idaho session law, 2026.

    statute, signed 31 March 2026, effective 1 July 2027; enforced by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the therapist provision reaches only statements that "explicitly" claim professional care.

  17. [17] Iowa General Assembly. Senate File 2417 (enrolled), conversational AI services (new Iowa Code chapter 554J). Iowa enrolled bill and governor’s approval letter, 2026.

    statute, approved by the governor 2 May 2026; applies from 1 July 2027; enforced by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: we found no acts chapter number; 1 July 2027 is stated as an applicability date.

  18. [18] Nebraska Legislature. Legislative Bill 525 (slip law), Conversational Artificial Intelligence Safety Act (sections 12 to 18). Nebraska slip law, 2026.

    statute, approved 14 April 2026; the chatbot sections become operative 1 July 2027; enforced by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the chatbot act rides in a bill mostly about agricultural data privacy.

  19. [19] Office of the Governor of California. Veto message, Senate Bill 903 (mental health professionals: artificial intelligence). governor’s veto message, dated 30 September 2026, 2026.

    veto message, matched against the Legislature’s official veto-message record; the bill history reads "Vetoed by the Governor" on 30 September 2026 · evidence tier: strong · funding: n/a (government) · accessed October 25, 2026

    the catch: a veto is not a ruling on the merits; the governor invited the author to return next year.

  20. [20] Massachusetts Legislature. S.2632, An Act relative to the use of artificial intelligence and other software tools in healthcare decision-making (194th General Court). pending bill, 2026.

    pending bill; reported favourably and referred to Senate Ways and Means on 2 April 2026; the same therapy language offered as a floor amendment to S.3178 was rejected on 23 July 2026 · evidence tier: early signal · funding: n/a (legislature) · accessed October 25, 2026

    the catch: not law; may not move before the session ends.

  21. [21] Pennsylvania General Assembly. Senate Bill 1090 (PN 1423), Safeguarding Adolescents from Exploitative Chatbots and Harmful AI Technology Act. pending bill, 2026.

    pending bill; passed the Senate 49-1 on 17 March 2026, referred to the House Communications and Technology committee · evidence tier: early signal · funding: n/a (legislature) · accessed October 25, 2026

    the catch: not law.

  22. [22] Pennsylvania General Assembly. House Bill 2100 (PN 2700), Protecting Patients Act. pending bill, 2025.

    pending bill; referred to House Professional Licensure on 12 December 2025, no action since · evidence tier: early signal · funding: n/a (legislature) · accessed October 25, 2026

    the catch: not law.

  23. [23] Texas Legislature. House Bill 149, Texas Responsible Artificial Intelligence Governance Act (enrolled; Business and Commerce Code chapter 552). Texas session law, 2025.

    statute, signed 22 June 2025, effective 1 January 2026; enforced exclusively by the attorney general · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: a general ai law; only its self-harm and health-care disclosure provisions are relevant here.

  24. [24] New York State Legislature. S.9008-C/A.10008-C (Chapter 58 of the Laws of 2026), Part Y, Safe by Design Act (General Business Law article 45-B). New York session law, 2026.

    statute (budget bill), shown as signed on 26 May 2026 in the Assembly’s action list; Part Y effective 1 January 2027 · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: reaches only ai companions built into social or gaming platforms, not standalone companion apps; signature read on the Assembly’s list, not the governor’s site.

  25. [25] Maine Legislature. Public Law 2025, chapter 294 (LD 1727 / HP 1154), An Act to Ensure Transparency in Consumer Transactions Involving Artificial Intelligence. Maine public law, 2025.

    statute, approved 12 June 2025; a violation is a violation of the Maine Unfair Trade Practices Act · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: a general disclosure rule for chatbots in trade and commerce; nothing about mental health; the text states no effective date.

  26. [26] New Hampshire General Court. HB 143 (Chapter 270, Laws of 2025), responsive generative communication to a child. New Hampshire session law, 2025.

    statute, approved 1 August 2025, effective 1 January 2026 · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: the title promises a private right of action but the text gives the attorney general "the sole right of action"; read it as unsettled.

  27. [27] California Legislature. Assembly Bill 1979, health care services: artificial intelligence (Chapter 854, Statutes of 2026). California statutes, 2026.

    statute, approved 30 September 2026; no operative date in the text, so 1 January 2027 by default · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: aimed at clinical decision support in health facilities; its chatbot piece works through california’s medical-confidentiality law.

  28. [28] California Legislature. Senate Bill 867, toys: companion chatbots (Chapter 189, Statutes of 2026). California statutes, 2026.

    statute, approved 10 September 2026; sunsets 1 January 2031 · evidence tier: strong · funding: n/a (statute) · accessed October 25, 2026

    the catch: about physical toys for children under 16, not apps.

  29. [29] US Federal Trade Commission. FTC Launches Inquiry into AI Chatbots Acting as Companions. FTC press release, 11 September 2025, 2025.

    section 6(b) orders to seven companies, approved 3-0 · evidence tier: moderate · funding: n/a (regulator) · accessed October 25, 2026

    the catch: a study without a specific law-enforcement purpose; no report published at our last check.

  30. [30] US Food and Drug Administration, Center for Devices and Radiological Health. Executive Summary for the Digital Health Advisory Committee Meeting: Generative Artificial Intelligence-Enabled Digital Mental Health Medical Devices, November 6, 2025. FDA advisory committee record, 2025.

    regulator briefing document prepared for the meeting · evidence tier: moderate · funding: n/a (regulator) · accessed October 25, 2026

    the catch: for discussion only; FDA says it is not draft or final guidance.

questions

Which states ban AI therapy?

Four states prohibit offering therapy or mental health services through AI unless a licensed professional provides them: Illinois (since 1 August 2025), Nevada (since 1 July 2025, which also bars AI from claiming to be a therapist), Rhode Island and Vermont (both since June 2026). Tennessee bars advertising an AI system as a qualified mental health professional (since 1 July 2026). Utah does not ban mental-health chatbots; it requires them to disclose that they are AI and bars selling users’ health information. Several companion-chatbot laws (Hawaii, Colorado, Idaho, Iowa, Nebraska, and Connecticut for minors) also bar chatbots from claiming to provide professional mental health care. Checked 25 October 2026.

Is it illegal for me to use ChatGPT or another chatbot for therapy?

No. None of these laws restricts what a person types into a chatbot. They bind the companies that build and advertise chatbots, and the licensed clinicians who might use AI in their practice.

What do companion-chatbot laws require?

The common core, starting with New York (in force 5 November 2025) and California SB 243 (1 January 2026): a protocol that detects expressions of suicidal ideation or self-harm and refers the user to crisis services such as 988 or a crisis text line, and a notice that the user is talking to an AI. New York requires the notice at the start of an interaction and at least every three hours, with attorney-general penalties up to $15,000 per day. Nine more states (Hawaii, Washington, Connecticut, Colorado, Oregon, Rhode Island, Idaho, Iowa and Nebraska) passed versions in 2026, most taking effect in 2027, and California added SB 1119 for children.

Can I sue a chatbot company under these laws?

Only under some. California SB 243 lets a person injured by a violation sue for the greater of actual damages or $1,000 per violation; Oregon’s SB 1546 does the same from 2027; California SB 1119 gives a narrower claim to a harmed child or parent; Vermont’s AI-therapy act gives private parties consumer-protection remedies. Hawaii, Connecticut, Idaho, Iowa and Nebraska say expressly that their laws create no private right of action. Illinois, Nevada, Utah, New York and Rhode Island leave enforcement to a state agency or the attorney general.

Did California ban AI therapy?

No. SB 903, which would have prohibited companion chatbots in psychotherapy and restricted clinicians’ use of AI, was vetoed on 30 September 2026. The governor wrote that the bill was overly broad and that its definition of psychotherapy services would capture general-purpose AI systems. California’s laws in force regulate companion chatbots (SB 243, and SB 1119 for children from 2027) rather than banning AI therapy.

Is there a federal law on AI therapy?

Not as of 25 October 2026. The Federal Trade Commission opened a 6(b) study of AI companion chatbots on 11 September 2025, sending orders to seven companies, but a 6(b) study has no specific law-enforcement purpose and no report had been published at our last check. FDA has authorised no AI-enabled device for a mental health use.

Do these laws apply to peer support?

Illinois and Rhode Island exempt peer support by name. Illinois defines it as services provided by individuals with lived experience of mental health conditions or recovery from substance use that are intended to offer encouragement, understanding, and guidance without clinical intervention. The companion-chatbot laws regulate AI systems, not people talking to people.

more on ai and mental health

  • is it ok to use chatgpt as a therapist? — the hub: the one randomised trial, the two pooled estimates that disagree, the stanford and rand tests, fda, the apa and three state laws.
  • red flags in a mental-health chatbot — twelve red flags and five green ones, each tied to a statute, the fda committee record or a published test, with the question to ask.
  • ai chatbots and teens’ mental health — for parents: one in five young people has asked a chatbot for mental-health advice and most told no one; what the studies, the ftc and the new state laws say.
  • ai chatbot lawsuits: the court record — one ruling at the motion-to-dismiss stage, settlements dismissed in october 2026, a chapter 11 case, and seven state suits against openai, from the dockets and the attorneys general’s own filings.

adam

the laws exempt people talking to people. that is all this is

veisund is free peer support, and you are anonymous to everyone you talk to. you post under a handle you pick — no real name, no phone number, nothing that follows you back to your real life 🤍

get veisund — it's free