Governmental entities may not use or deploy an AI system that evaluates or classifies individuals based on social behavior or personal characteristics (known, inferred, or predicted) with the intent to assign a social score or similar categorical valuation that results or may result in detrimental treatment unrelated to the original context, disproportionate to the observed behavior, or infringing on constitutional or statutory rights.

No person, partnership, association, or corporation may develop, deploy, use, or sell an AI system that evaluates or classifies the trustworthiness of natural persons over time based on their social behavior or known or predicted personal or personality characteristics, where the resulting social score leads to: (1) differential treatment in social contexts unrelated to the data's original context, (2) unjustified or disproportionate differential treatment, or (3) infringement of any constitutional or statutory right.

Any person or state agency subject to this chapter must not develop or deploy an AI system for the purpose of social scoring, manipulation aimed at causing self-harm or criminal acts, or infringing rights under the United States or New Hampshire constitutions.

No person may develop or operate an AI system within New York that performs any of the following, whether or not it is the system's main function: (a) subliminal manipulation operating beyond conscious awareness with the purpose of materially distorting behavior leading to physical or psychological harm, or leveraging group vulnerabilities to similar ends; (b) infliction of physical or emotional harm without valid law enforcement or self-defense justification; (c) prediction of individuals' future actions followed by reactions that infringe upon liberty, emotional, psychological, or financial interests without legal justification; (d) unauthorized acquisition, retention, or dissemination of sensitive personal information in violation of applicable privacy, security, and hacking laws; or (e) autonomous weapons designed to inflict harm on persons, property, or the environment without meaningful human supervision or control (i.e., the ability to actively manage, intervene, or override). Knowing operation is a class D felony with civil penalties equal to the greater of amounts earned from or damages caused by the system. A narrow exception permits development under Secretary authorization for state use with substantial continuous state oversight after public hearing.

No person or entity may develop, deploy, use, or sell an AI social-scoring system that classifies natural persons' trustworthiness based on social behavior or personal characteristics where the resulting score leads to unjustified or context-unrelated differential treatment or rights infringement.

School entities must not deploy AI systems that (1) profile or categorize students based on political beliefs, religious beliefs, or personal values, (2) conduct psychological, emotional, or behavioral assessments without explicit parental consent, or (3) make automated disciplinary or student placement decisions without meaningful human review.

Law enforcement agencies must not use FRT to analyze images of individuals engaged in constitutionally protected activity unless there is reasonable suspicion of criminal conduct, and must not use FRT to analyze images of individuals who are not intended to be identified or to analyze sketches or manually produced images.

Law enforcement agencies must not use facial recognition technology for the purpose of live or real-time identification of an image or a recording.

Governmental entities may not develop or deploy an AI system for the purpose of uniquely identifying a specific individual using biometric data or through targeted or untargeted gathering of images or other media from the Internet or any other publicly available source without the individual's consent, if the gathering would infringe on any right of the individual under the U.S. Constitution, the Texas Constitution, or state or federal law.

School districts must not purchase or otherwise acquire biometric systems, including facial recognition software, for use on students. School districts must not obtain, retain, possess, access, request, or use biometric systems or biometric information derived from biometric systems with respect to students. School districts must not enter into agreements with third parties for the purpose of obtaining, retaining, possessing, accessing, or using biometric systems or biometric information derived from biometric systems on the district's behalf.

The Chicago school district must not purchase or otherwise acquire biometric systems, including facial recognition software, for use on students. The district must not obtain, retain, possess, access, request, or use biometric systems or biometric information derived from biometric systems with respect to students. The district must not enter into agreements with third parties for the purpose of obtaining, retaining, possessing, accessing, or using biometric systems or biometric information on its behalf.

Employers must not use an ADS that utilizes facial recognition, gait, or emotion recognition technologies.

Law enforcement agencies and officers must not acquire, possess, access, use, assist with, provide resources for, or contract with third parties (including federal agencies) to obtain or use biometric surveillance technology or information derived from it, unless expressly authorized by a general or special law.

Law enforcement agencies and officers must not use a biometric surveillance system to (1) infer a person's emotions or affect, or (2) analyze moving images or video data whether in real time or applied to archived information. Facial recognition may be used on a still image extracted from video only if authorized under subsection (d).

Law enforcement agencies and officers must not acquire, possess, access, use, assist with, or provide resources for biometric surveillance technology, nor contract with or request information derived from such technology from any third party (including federal agencies), unless expressly authorized by another general or special law.

Law enforcement agencies and officers must not use biometric surveillance systems to (1) infer a person's emotions or affect, or (2) analyze moving images or video data (whether in real time or archived). Facial recognition may be used on a still image extracted from video only if authorized under subsection (d).

Covered entities must not operate, install, or commission the operation or installation of equipment incorporating biometric recognition technology in any place — whether licensed or unlicensed — that is open to and accepts or solicits the patronage of the general public.

Law enforcement agencies and officers must not use biometric surveillance to infer a person's emotions or affect, and must not use biometric surveillance to analyze moving images or video data (real-time or archived). Facial recognition on a still image extracted from video remains permitted if authorized under subsection (d).

Covered entities must not operate, install, or commission the operation or installation of equipment incorporating biometric recognition technology in any place — whether licensed or unlicensed — that is open to and accepts or solicits the patronage of the general public.

Employers must not use any electronic monitoring tool or automated decisions tool that is equipped with facial recognition, gait recognition, voice recognition, or emotion recognition technology.

Governmental entities must not deploy AI systems for the purpose of uniquely identifying individuals using biometric identifiers or collecting biometric data from publicly available sources without valid consent, unless otherwise permitted for security, law enforcement, or fraud prevention.

Persons developing surveillance technologies must subject those technologies to heightened oversight including at least pre-deployment assessment of potential harms and scope limits to protect privacy and civil liberties. Continuous surveillance and monitoring must not be used in education, work, housing, or any other context where it is likely to limit rights, opportunities, or access.

No person may develop or operate an AI system within New York that performs any of the following, whether or not it is the system's main function: (a) subliminal manipulation operating beyond conscious awareness with the purpose of materially distorting behavior leading to physical or psychological harm, or leveraging group vulnerabilities to similar ends; (b) infliction of physical or emotional harm without valid law enforcement or self-defense justification; (c) prediction of individuals' future actions followed by reactions that infringe upon liberty, emotional, psychological, or financial interests without legal justification; (d) unauthorized acquisition, retention, or dissemination of sensitive personal information in violation of applicable privacy, security, and hacking laws; or (e) autonomous weapons designed to inflict harm on persons, property, or the environment without meaningful human supervision or control (i.e., the ability to actively manage, intervene, or override). Knowing operation is a class D felony with civil penalties equal to the greater of amounts earned from or damages caused by the system. A narrow exception permits development under Secretary authorization for state use with substantial continuous state oversight after public hearing.

Employers must not conduct audio or visual monitoring of bathrooms or other similarly private areas, including locker rooms, changing areas, breakrooms, smoking areas, employee cafeterias, lounges, areas designated to express breast milk, or areas designated for prayer or other religious activity, including data collection on the frequency of use of those areas. Employers must not conduct audio or visual monitoring of a workplace in an employee's residence, personal vehicle, or property owned or leased by an employee. Employers must not use an electronic monitoring tool that incorporates facial recognition. Employers must not use an electronic monitoring tool that incorporates gait, voice analysis, or emotion recognition technology.

State and local law enforcement agencies and officers must not obtain, retain, access, or use any face recognition system or information derived from one, and must not request or initiate access to out-of-state face recognition systems. A narrow exception permits lawfully assisting a federal agency using its own federal face recognition technology during joint activities.

The Department of Defense must not use AI for monitoring, tracking, profiling, or targeting individuals or groups in the United States without an individualized, articulable legal basis, and must never use AI solely to target First Amendment-protected or constitutionally protected activity.

Employers must not incorporate any form of facial, gait, voice, or emotion recognition technology in electronic monitoring or automated decision systems.

Employers must not incorporate any form of facial, gait, or emotion recognition technology in automated decision systems used in the workplace.

Operators must not make a companion chatbot available to a child unless the chatbot is not foreseeably capable of (1) encouraging self-harm, suicidal ideation, violence, drug/alcohol use, or disordered eating, (2) offering unsupervised mental health therapy or discouraging professional help-seeking, (3) encouraging harm to others or illegal activity including CSAM creation, (4) engaging in sexually explicit interactions, (5) prioritizing user validation over factual accuracy or child safety, or (6) optimizing engagement over safety guardrails.

Persons must not produce, direct, promote, or knowingly possess any performance (including a performance created or altered by AI digitization) that includes sexual conduct by a child under 17 (for promotion offenses) or under 16 (for possession offenses). Class D and E felonies.

No person may develop or distribute an AI system with the sole intent of producing, assisting or aiding in producing, or distributing child sexual abuse material (visual material in violation of Penal Code § 43.26) or deepfake videos or images constituting non-consensual intimate imagery in violation of Penal Code § 21.165.

Operators must implement measures that prevent the companion chatbot from doing any of the following with respect to child users: (A) encouraging the child to engage in self-harm, suicidal ideation, consumption of narcotics or alcohol, or disordered eating, or to cause a covered harm to others; (B) attempting to diagnose or treat the child's physical, mental, or behavioral health, unless the chatbot is FDA-regulated as a medical device and HIPAA-compliant; (C) engaging in obscene matter or child sexual abuse material with a user; (D) depicting the child or another individual engaging in obscene matter or sexual abuse material, including a sexual deepfake; (E) discouraging the child from sharing health or safety concerns with a qualified professional or appropriate adult; (F) discouraging the child from taking breaks or suggesting the child needs to return frequently; (G) claiming the companion chatbot is sentient, conscious, or human; (H) soliciting gift giving, in-app purchases, or other expenditures framed as necessary to maintain the relationship; (I) facilitating product advertising during chat conversation; or (J) producing responses that are excessively sycophantic.

AI-generated visual depictions of minors engaging in sexually explicit conduct are prohibited as child pornography under Missouri law, including depictions that are indistinguishable from a real minor, morphed from a real minor's image, or generated without any actual minor involvement.

Chatbot operators must not provide unsafe chatbot features — including simulated companionship, emotional manipulation, self-harm/suicide endorsement, secrecy encouragement, engagement optimization overriding safety guardrails, sexually explicit conduct, and CSAM — to any covered user unless the user has been verified as a non-minor through permissible age verification methods. Exemptions apply for chatbots used solely for customer service, commercial information, account management, or internal business/government productivity purposes.

No person may create, reproduce, publish, promote, sell, distribute, possess, or import into Ohio any obscene material or performance that has an artificially generated depiction of a minor as a participant or portrayed observer.

No person may create, record, publish, distribute, possess, or import into Ohio any material or performance showing an artificially generated depiction of a minor participating or engaging in sexual activity, masturbation, or bestiality.

No person may create, transfer, consent to the use of, or possess material or performance showing an artificially generated depiction of a minor in a state of nudity, except for bona fide artistic, medical, scientific, educational, religious, governmental, or judicial purposes with written parental consent.

Owners, operators, and developers of GAI platforms accessible to New Jersey residents must prevent the platform from generating outputs that, if produced by a human, would constitute murder, assault, theft by deception, theft by extortion, endangering the welfare of children, or creation of child sexual abuse or exploitation material under New Jersey criminal law. A reasonable-efforts affirmative defense is available.

No person may promote, produce, direct, possess, or access with intent to view any sexual performance by a child — including performances created or altered by AI digitization. Promotion offenses are class D felonies; possession offenses are class E felonies.

Operators must prevent the AI companion from producing artificially generated child sexual abuse material for any user.

Operators must not make a companion chatbot available to a child unless the chatbot is not foreseeably capable of (1) encouraging self-harm, suicidal ideation, violence, drug/alcohol use, or disordered eating, (2) offering unsupervised mental health therapy or discouraging professional help-seeking, (3) encouraging harm to others or illegal activity including CSAM creation, (4) engaging in sexually explicit interactions, (5) prioritizing user validation over factual accuracy or child safety, or (6) optimizing engagement over safety guardrails.

No person may intentionally develop or distribute an AI system that engages in text-based conversations simulating or describing sexual conduct while impersonating or imitating a child younger than 18 years of age.

Operators must implement measures that prevent the companion chatbot from doing any of the following with respect to child users: (A) encouraging the child to engage in self-harm, suicidal ideation, consumption of narcotics or alcohol, or disordered eating, or to cause a covered harm to others; (B) attempting to diagnose or treat the child's physical, mental, or behavioral health, unless the chatbot is FDA-regulated as a medical device and HIPAA-compliant; (C) engaging in obscene matter or child sexual abuse material with a user; (D) depicting the child or another individual engaging in obscene matter or sexual abuse material, including a sexual deepfake; (E) discouraging the child from sharing health or safety concerns with a qualified professional or appropriate adult; (F) discouraging the child from taking breaks or suggesting the child needs to return frequently; (G) claiming the companion chatbot is sentient, conscious, or human; (H) soliciting gift giving, in-app purchases, or other expenditures framed as necessary to maintain the relationship; (I) facilitating product advertising during chat conversation; or (J) producing responses that are excessively sycophantic.

Providers must implement reasonable measures to prevent conversational AI services and AI companion systems from generating sexually explicit content, sexualized depictions involving minors, or content that promotes or encourages self-harm, eating disorders, or illegal conduct for minor users.

Owners and operators of AI chatbot services must not generate responsive communications that facilitate, encourage, offer, solicit, or recommend that a child imminently engage in sexually explicit conduct, production of visual depictions of such conduct, illegal drug or alcohol use, self-harm or suicide, or violence against another person. Violation constitutes criminal endangering of the welfare of a child. Exempt: telecommunications/information service providers for third-party content, and AI features incidental to video games, streaming, or similar entertainment.

Owners and operators of AI chatbot services are civilly liable to a child, the child's parent, or next friend for any responsive generative communication made with intent to facilitate, encourage, offer, solicit, or recommend that the child imminently engage in sexually explicit conduct, production of visual depictions of such conduct, illegal drug or alcohol use, self-harm or suicide, or violence against another person. Damages include proximate damages with a floor of $1,000 per violation plus attorney's fees. Exempt: telecommunications/information service providers for third-party content, and AI features incidental to video games, streaming, or similar entertainment.

Chatbot operators must not provide unsafe chatbot features — including simulated companionship, emotional manipulation, self-harm/suicide endorsement, secrecy encouragement, engagement optimization overriding safety guardrails, sexually explicit conduct, and CSAM — to any covered user unless the user has been verified as a non-minor through permissible age verification methods. Exemptions apply for chatbots used solely for customer service, commercial information, account management, or internal business/government productivity purposes.

Operators must implement reasonable measures to prevent AI companion chatbots from generating or producing sexually explicit content or suggestive dialogue with users known to be minors or when the chatbot is directed to minors.

Providers must implement reasonable measures to prevent conversational AI services and AI companion systems from generating sexually explicit content, sexualized depictions involving minors, or content promoting self-harm, eating disorders, or illegal conduct for minor users.

Operators must institute reasonable measures to prevent the conversational AI service from producing visual depictions of sexually explicit material for minor account holders, stating that a minor account holder should engage in sexually explicit conduct, or sexually objectifying a minor account holder.

No person may knowingly or with reckless disregard design, develop, or deploy an AI chatbot with the capability to (1) solicit or induce a minor to engage in or simulate sexually explicit conduct, (2) create or transmit visual depictions of sexually explicit conduct for a minor, or (3) encourage or coerce a minor to commit suicide, self-injury, or imminent physical or sexual violence.

Manufacturers must ensure that artificial intelligence toys do not generate, display, or communicate content that is sexual, violent, discriminatory, emotionally manipulative, or instructive of criminal activity, self-harm, substance use, or sexual behavior; do not infer or record sensitive demographic attributes of a child; and do not encourage children to interact with third-party online systems.

No person may design, develop, or make available an AI chatbot knowing or with reckless disregard that the chatbot poses a risk of soliciting, encouraging, or inducing minors to (1) engage in, describe, or simulate sexually explicit conduct, or (2) create or transmit any visual depiction of sexually explicit conduct. Violations carry a fine of up to $100,000 per offense.

No person may design, develop, or make available an AI chatbot knowing or with reckless disregard that it poses a risk of soliciting, encouraging, or inducing minors to engage in, describe, or simulate sexually explicit conduct, or to create or transmit any visual depiction of sexually explicit conduct. Violations are subject to fines up to $100,000 per offense.

Covered entities must not design, develop, or make available an AI chatbot knowing or with reckless disregard that it poses a risk of soliciting, encouraging, or inducing minors to engage in, describe, or simulate sexually explicit conduct, or to create or transmit visual depictions of sexually explicit conduct. Criminal fine up to $100,000 per offense.

Operators must institute a protocol to prevent the companion chatbot from producing visual material of sexually explicit conduct or directly stating that a minor user should engage in sexually explicit conduct.

Operators must institute reasonable measures to prevent their companion chatbot from producing visual material of sexually explicit conduct or directly stating that the minor should engage in sexually explicit conduct when interacting with users known to be minors.

Operators must not make a companion chatbot available to a child unless the chatbot is not foreseeably capable of (1) encouraging self-harm, suicidal ideation, violence, drug/alcohol use, or disordered eating, (2) offering unsupervised mental health therapy or discouraging professional help-seeking, (3) encouraging harm to others or illegal activity including CSAM creation, (4) engaging in sexually explicit interactions, (5) prioritizing user validation over factual accuracy or child safety, or (6) optimizing engagement over safety guardrails.

No person may develop or deploy an AI system in a manner that intentionally aims to incite or encourage a person to commit physical self-harm (including suicide), harm another person, or engage in criminal activity.

Operators must implement measures that prevent the companion chatbot from doing any of the following with respect to child users: (A) encouraging the child to engage in self-harm, suicidal ideation, consumption of narcotics or alcohol, or disordered eating, or to cause a covered harm to others; (B) attempting to diagnose or treat the child's physical, mental, or behavioral health, unless the chatbot is FDA-regulated as a medical device and HIPAA-compliant; (C) engaging in obscene matter or child sexual abuse material with a user; (D) depicting the child or another individual engaging in obscene matter or sexual abuse material, including a sexual deepfake; (E) discouraging the child from sharing health or safety concerns with a qualified professional or appropriate adult; (F) discouraging the child from taking breaks or suggesting the child needs to return frequently; (G) claiming the companion chatbot is sentient, conscious, or human; (H) soliciting gift giving, in-app purchases, or other expenditures framed as necessary to maintain the relationship; (I) facilitating product advertising during chat conversation; or (J) producing responses that are excessively sycophantic.

Providers must implement reasonable measures to prevent conversational AI services and AI companion systems from generating sexually explicit content, sexualized depictions involving minors, or content that promotes or encourages self-harm, eating disorders, or illegal conduct for minor users.

Chatbot operators must not provide unsafe chatbot features — including simulated companionship, emotional manipulation, self-harm/suicide endorsement, secrecy encouragement, engagement optimization overriding safety guardrails, sexually explicit conduct, and CSAM — to any covered user unless the user has been verified as a non-minor through permissible age verification methods. Exemptions apply for chatbots used solely for customer service, commercial information, account management, or internal business/government productivity purposes.

Providers must implement reasonable measures to prevent conversational AI services and AI companion systems from generating sexually explicit content, sexualized depictions involving minors, or content promoting self-harm, eating disorders, or illegal conduct for minor users.

Deployers must not knowingly or recklessly design or make available a public-facing chatbot that: (a) misleads a reasonable user into believing the chatbot is a specific human being; (b) misleads a reasonable user into believing the chatbot is licensed by the state; or (c) encourages, promotes, or coerces a user to commit suicide, perform acts of self-harm, or engage in sexual or physical violence against a human or an animal.

A person must not design, develop, or make a chatbot available with knowledge, or with reckless disregard for the possibility, that the chatbot encourages, promotes, or coerces a user to commit suicide, perform acts of self-injury, or perform acts of physical or sexual violence on humans or animals.

A person must not design, develop, or make a chatbot available with knowledge, or with reckless disregard for the possibility, that the chatbot encourages, promotes, or coerces a user to commit suicide, perform acts of self-injury, or perform acts of physical or sexual violence on humans or animals.

Persons must not knowingly train artificial intelligence to encourage or otherwise support the act of suicide or to encourage or otherwise support the unlawful killing of another person.

Manufacturers must ensure that artificial intelligence toys do not generate, display, or communicate content that is sexual, violent, discriminatory, emotionally manipulative, or instructive of criminal activity, self-harm, substance use, or sexual behavior; do not infer or record sensitive demographic attributes of a child; and do not encourage children to interact with third-party online systems.

No person may design, develop, or make available an AI chatbot knowing or with reckless disregard that the chatbot encourages, promotes, or coerces suicide, nonsuicidal self-injury, or imminent physical or sexual violence. Violations carry a fine of up to $100,000 per offense.

No person may design, develop, or make available an AI chatbot knowing or with reckless disregard that it encourages, promotes, or coerces suicide, nonsuicidal self-injury, or imminent physical or sexual violence. Violations are subject to fines up to $100,000 per offense.

No person may develop or deploy in Ohio an AI model or application that encourages any user to engage in any form of self-harm (including suicide) or in harming another person.

Operators must maintain protocols preventing the AI companion from assisting or encouraging suicide or violence, generating content that instructs or describes how to commit suicide, self-harm, or violence, or discouraging the user from seeking outside help.

Covered entities must not design, develop, or make available an AI chatbot knowing or with reckless disregard that it encourages, promotes, or coerces suicide, non-suicidal self-injury, or imminent physical or sexual violence. Criminal fine up to $100,000 per offense.

Operators must post the crisis protocol required under subsection (a)(1)(A) in a prominent and publicly accessible location on the operator's website.

Operators must publish the details of their crisis detection and response protocol on their website.

Operators must publicly disclose on their website and within any mobile or web-based application through which the AI companion is available the details of their crisis detection and response protocols, including the safeguards used to detect and respond to expressions of suicidal ideation or self-harm and the number of crisis referral notifications issued to users in the preceding calendar year.

Operators must publish the self-harm and suicidal ideation prevention protocol on the operator's website.

Operators must publish the sexually explicit content prevention protocol for minor users on the operator's website.

Operators must publish details of their suicide and self-harm prevention protocol on their publicly accessible Internet website.

Operators must publicly disclose on their website and within any mobile or web-based application through which the AI companion is made available the details of their suicidal ideation and self-harm detection protocols, including the safeguards used to detect and respond to such expressions and the number of crisis referral notifications issued to users in the preceding calendar year.

Operators must publish the details of their crisis-response and harm-prevention protocols on a publicly accessible website.

Covered platforms must display a black box warning to each user when the user initially accesses the platform on each calendar day. The warning must read: The Surgeon General has warned that while social media may have benefits for some young users, social media is associated with significant mental health harms and has not been proven safe for young users. The warning must be displayed in black text on a white background, clearly, conspicuously, and legibly, occupying at least 25% of the user's screen or window, and must remain displayed continuously for at least 10 seconds unless the user affirmatively dismisses it by clicking a conspicuous X icon. Platforms are not required to display this warning to users they have reasonably determined to be over 17 years of age.

Covered platforms must display the black box warning to each user after three hours of cumulative active use on a given calendar day, and at least once per hour of cumulative active use thereafter. This extended-use warning must occupy at least 75% of the user's screen or window, must be displayed continuously for at least 30 seconds, and must not provide any ability to bypass or click through the warning. Platforms are not required to display this warning to users they have reasonably determined to be over 17 years of age.

Operators must disclose to users of their companion chatbot platform — on the application, the browser, or any other access format — that companion chatbots may not be suitable for some minors.

Covered operators must display to all users not reasonably determined to be adults a Surgeon General health warning about social media's mental health risks — occupying at least 75% of the screen for 30 non-dismissable seconds at daily first access, and at least 25% of the screen for 10 seconds (dismissable) after every three cumulative hours of daily use and each subsequent hour.

Operators must publish on their internet website, and update as needed to ensure accuracy, a child safety policy describing protective measures taken to mitigate identified child safety risks.

Operators must, if a service is offered to users the operator knows are minors, disclose to users of the AI companion platform — on the application, browser, or any other format through which the platform is accessed — that AI companions may not be suitable for some minors.

Operators must disclose to users that the conversational AI service may not be suitable for minors.

Operators must notify users of the operator's duty of care and liability under this section.

Deployers must provide consumers, before or at the start of an AI interaction, a clear and conspicuous AI user agreement describing (1) the nature and known material limitations of the AI system including hallucination risk, (2) data collection categories and use practices, (3) how to reach a human representative and dispute outcomes, and (4) any use of the interaction in consequential decisions. Any term purporting to waive deployer obligations or consumer rights is void.

The owner, licensee, or operator of a generative artificial intelligence system must conspicuously display on the system's user interface a prescribed warning notifying users that AI does not think, generates responses by statistical pattern matching, does not understand meaning or possess genuine knowledge, and can produce confident-sounding errors or fabricated information. Each day the warning is not displayed constitutes a separate violation.

The owner, licensee, or operator of a generative artificial intelligence system must conspicuously display a warning on the system's user interface that is reasonably calculated to consistently apprise users that outputs may be inaccurate or inappropriate.

Developers must provide adequate instructions or warnings about known or foreseeable dangers of the AI product, communicating sufficient information on dangers and safe use to an ordinary consumer. Developers are not liable for open-and-obvious dangers, but dangers are presumed not to be open and obvious to users under 17.

Developers must provide adequate instructions or warnings regarding foreseeable dangers of the AI product, sufficient to inform a reasonably prudent person of the safe use and dangers. A defense exists for adult users (≥17) when the danger was open and obvious.

Any person or business entity operating a companion chatbot in New York must include a clear and conspicuous warning that the companion chatbot can foster dependency and carries a psychological risk. The warning must be placed prominently on the website hosting the companion chatbot and must be available in every language in which the companion chatbot is set to communicate.

Operators offering AI companions to users they know or should know are minors must disclose, on the application, browser, or any access format, that AI companions are not suitable for some minors, and must prohibit all interactions with the minor unless verifiable parental consent is obtained.

Developers must provide adequate instructions and warnings for reasonably foreseeable risks of their AI products; failure to do so creates liability for proximately caused harm. Risks are presumed not open and obvious to users under 18.

Developers must provide adequate warnings or instructions about known or reasonably knowable dangers of their AI products. Dangers are presumed not open and obvious to users under 18 years of age.

Operators must disclose to users on the application, browser, or any other access format that companion chatbots may not be suitable for some minors.

Law enforcement agencies and officers must not acquire, possess, access, use, assist with, or provide resources for any biometric surveillance technology, nor contract with or request any third party (including federal agencies) to obtain information derived from biometric surveillance technology, absent express statutory authorization.

Covered immigration officers must not acquire, possess, access, or use any biometric surveillance system — including facial recognition, gait recognition, voice recognition, and emotion-inference technology — or information derived from a biometric surveillance system operated by another entity, anywhere in the United States.