WHAT THIS BILL REGULATES · 1 REQUIREMENT TYPE
How Is This Bill Enforced
Verbatim statutory text on the left; plain-language analysis and a per-section checklist on the right. Numbered markers cross-link to the matching checklist row.
(1)(a)–(c) "Visual or audio generative artificial intelligence systemVisual or audio generative artificial intelligence system"Visual or audio generative artificial intelligence system" or "covered system" shall mean any artificial intelligence system that is accessible to New York residents whose primary function is to generate visual or auditory media.Gen. Bus. Law § 390-f(1)(a)" or "covered system" shall mean any artificial intelligenceArtificial intelligence"Artificial intelligence" shall mean a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. Artificial intelligence systems use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action. This definition includes but is not limited to systems that use machine learning, large language models, natural language processing, and computer vision technologies, including generative artificial intelligence.Gen. Bus. Law § 390-f(1)(b) system that is accessible to New York residents whose primary function is to generate visual or auditory media. (b) "Artificial intelligenceArtificial intelligence"Artificial intelligence" shall mean a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. Artificial intelligence systems use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action. This definition includes but is not limited to systems that use machine learning, large language models, natural language processing, and computer vision technologies, including generative artificial intelligence.Gen. Bus. Law § 390-f(1)(b)" shall mean a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. Artificial intelligenceArtificial intelligence"Artificial intelligence" shall mean a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. Artificial intelligence systems use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action. This definition includes but is not limited to systems that use machine learning, large language models, natural language processing, and computer vision technologies, including generative artificial intelligence.Gen. Bus. Law § 390-f(1)(b) systems use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action. This definition includes but is not limited to systems that use machine learning, large language models, natural language processing, and computer vision technologies, including generative artificial intelligenceArtificial intelligence"Artificial intelligence" shall mean a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. Artificial intelligence systems use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action. This definition includes but is not limited to systems that use machine learning, large language models, natural language processing, and computer vision technologies, including generative artificial intelligence.Gen. Bus. Law § 390-f(1)(b). (c) "Covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c)" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) shall include the covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c)'s agent, employee or representative acting at the direction of the covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c).
Subdivision 1 establishes the definitional framework for the bill. It defines three key terms: the covered system (visual or audio generative AI accessible to New York residents), artificial intelligence broadly, and covered person (public officeholders and candidates). The covered-person definition is notably narrow — it protects only public officials and candidates, not private individuals. The covered-system definition is broad, encompassing any AI system whose primary function is generating visual or auditory media.
(2) 1 The owner, licensee or operator of a visual or audio generative artificial intelligence systemVisual or audio generative artificial intelligence system"Visual or audio generative artificial intelligence system" or "covered system" shall mean any artificial intelligence system that is accessible to New York residents whose primary function is to generate visual or auditory media.Gen. Bus. Law § 390-f(1)(a) shall implement a reasonable method to prohibit its users from creating unauthorized realistic depictions of a covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) within sixty days of being notified by such covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) that such covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) does not want a realistic depiction of themselves to be generated by the owner, licensee or operatorOwner, licensee or operatorThe owner, licensee or operator of a visual or audio generative artificial intelligence system (no formal statutory definition provided; term used throughout § 390-f to identify the regulated party).Gen. Bus. Law § 390-f(2)'s system. An implemented method to prevent the unauthorized creation of realistic depictions of a covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) shall be considered reasonable when the owner, licensee or operator of the covered system has implemented a method that, in relation to the method of user inputs used by the covered system, is consistent with industry standards, not overly burdensome on the system, cost-effective to implement and maintain and is up to date.
Subdivision 2 imposes the bill's core obligation: upon receiving notice from a covered person, the owner, licensee, or operator of a covered system must implement a reasonable method to prevent users from creating unauthorized realistic depictions of that covered person within 60 days. The statute defines reasonableness by reference to industry standards, system burden, cost-effectiveness, and currency — creating a flexible safe harbor rather than prescribing a specific technical approach.
(3) 2 The owner, licensee or operator of a covered system shall implement a reasonable method for covered personsCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) to send notice to them under this section provided that such method is easy to access, understand, complete and send and that such method provides clear updates to the sender on the status of their request in a timely manner. The owner, licensee or operator of a covered system may request a reasonable means of identification to process such requests.
Subdivision 3 requires operators to provide an accessible notice mechanism through which covered persons can submit their opt-out requests. The mechanism must be easy to access, understand, complete, and send, and must provide clear, timely status updates. Operators may request reasonable identification from the covered person to process the request. This creates a procedural counterpart to the substantive obligation in subdivision 2.
(4) Nothing in this section shall prohibit the owner, licensee or operator of a covered system from implementing reasonable safeguards to permit the covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c), their agent, employee or representative to use such covered system to generate realistic depictions of such covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c), provided however that such owner, licensee or operator of such covered system shall be liable in the same manner as if they had violated subdivision two of this section where such safeguards are not reasonable. A safeguard is considered reasonable for purposes of this subdivision where the owner, licensee or operator of a covered system implements measures that are consistent with industry standards, not overly burdensome on the system, cost-effective to implement and maintain and are up to date.
Subdivision 4 is a savings clause permitting operators to implement safeguards that allow the covered person (or their agents) to use the system to generate realistic depictions of themselves, even after submitting an opt-out notice. However, operators remain liable if those safeguards are not reasonable, applying the same reasonableness standard as subdivision 2 — consistency with industry standards, not overly burdensome, cost-effective, and current.
(5) 3 The owner, licensee or operator of a covered system shall be liable to a covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) in an amount of one hundred dollars per depiction, but not more than one hundred thousand dollars in the aggregate, generated on their system by a user other than the covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) or their agent, employee, representative, or another at the direction of them created outside of the period prescribed by this section where such owner, licensee or operator of such covered system fails to implement a reasonable method to prevent the unauthorized creation of realistic depictions of a covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) within the periods prescribed in this section. The owner, licensee or operator of a covered system shall not be liable to a covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) where such owner, licensee or operatorOwner, licensee or operatorThe owner, licensee or operator of a visual or audio generative artificial intelligence system (no formal statutory definition provided; term used throughout § 390-f to identify the regulated party).Gen. Bus. Law § 390-f(2) is unable to prevent the unauthorized depictions of such covered personCovered person"Covered person" shall mean any person holding a public office or a candidate for public office. In the context of notice, covered person shall include the covered person's agent, employee or representative acting at the direction of the covered person.Gen. Bus. Law § 390-f(1)(c) after implementing a reasonable method, nor shall they be liable for any unauthorized realistic depictions that are incidental or were created in an unforeseeable way.
Subdivision 5 establishes the damages framework. Operators face $100 per unauthorized depiction generated by a user other than the covered person (or their agents), up to an aggregate cap of $100,000. The liability only applies where the operator fails to implement a reasonable method within the prescribed timeframe. Two safe harbors limit liability: (1) operators are not liable where they implemented a reasonable method but were unable to prevent the depictions, and (2) operators are not liable for incidental or unforeseeable unauthorized depictions.
(6) This section shall not apply to the owner, licensee or operator of a visual or audio generative artificial intelligence systemVisual or audio generative artificial intelligence system"Visual or audio generative artificial intelligence system" or "covered system" shall mean any artificial intelligence system that is accessible to New York residents whose primary function is to generate visual or auditory media.Gen. Bus. Law § 390-f(1)(a) where the visual or audio outputs of the system are processed by a third party that has no ownership or control over the underlying generative model.
Subdivision 6 exempts operators where the visual or audio outputs of the system are processed by a third party with no ownership or control over the underlying generative model. This carve-out appears designed to shield operators when a downstream third party takes the system's raw output and processes it into a depiction, breaking the causal chain between the operator's system and the final unauthorized depiction.
This act shall take effect one year after it shall have become a law.
Section 2 provides that the act takes effect one year after it becomes law. Because the bill has not been enacted, the effective date is not yet determinable.