A As used in this Chapter, the following words shall mean the following: (1) "AffiliateAffiliate"Affiliate" means a person controlling, controlled by, or under common control with a specified person.R.S. 51:3111.2(A)(1)" means a person controlling, controlled by, or under common control with a specified person. (2) "Artificial intelligence modelArtificial intelligence model"Artificial intelligence model" means a machine-based system that receives inputs and generates outputs, including but not limited to predictions, content, recommendations, or decisions that can influence physical or virtual environments. "Artificial intelligence model" includes machine learning models, deep learning models, and other computational models designed to simulate aspects of human intelligence.R.S. 51:3111.2(A)(2)" means a machine-based system that receives inputs and generates outputs, including but not limited to predictions, content, recommendations, or decisions that can influence physical or virtual environments. "Artificial intelligence modelArtificial intelligence model"Artificial intelligence model" means a machine-based system that receives inputs and generates outputs, including but not limited to predictions, content, recommendations, or decisions that can influence physical or virtual environments. "Artificial intelligence model" includes machine learning models, deep learning models, and other computational models designed to simulate aspects of human intelligence.R.S. 51:3111.2(A)(2)" includes machine learning models, deep learning models, and other computational models designed to simulate aspects of human intelligence. (3)(a) "Catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3)" means a foreseeable and material risk that a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10)'s development, storage, use, or deployment of a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in propertyProperty"Property" means corporeal or incorporeal property.R.S. 51:3111.2(A)(15) damage or loss arising from a single incident involving a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11): (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. (b) "Catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3)" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation modelFoundation model"Foundation model" means an artificial intelligence model that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks.R.S. 51:3111.2(A)(8). (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) in combination with other software if the frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) did not materially contribute to the harm. (4) "Critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4)" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4)" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities. (5) "Critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5)" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) that results in death, bodily injury, or a cyberattack against critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4). (b) Harm resulting from the materialization of a catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) as defined in this Section. (c) Loss of control of a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) causing death, bodily injury, or damage to critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4). (d) A frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) that uses deceptive techniques against the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) to subvert the controls or monitoring of its frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3). (6) "DepartmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6)" means the Louisiana Department of Justice, office of the attorney general. (7) "DeployDeploy"Deploy" means to make a frontier model available to a third party, except for the primary purpose of developing or evaluating the frontier model.R.S. 51:3111.2(A)(7)" means to make a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) available to a third party, except for the primary purpose of developing or evaluating the frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11). (8) "Foundation modelFoundation model"Foundation model" means an artificial intelligence model that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks.R.S. 51:3111.2(A)(8)" means an artificial intelligence modelArtificial intelligence model"Artificial intelligence model" means a machine-based system that receives inputs and generates outputs, including but not limited to predictions, content, recommendations, or decisions that can influence physical or virtual environments. "Artificial intelligence model" includes machine learning models, deep learning models, and other computational models designed to simulate aspects of human intelligence.R.S. 51:3111.2(A)(2) that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks. (9) "Frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9)" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risksCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3), with specific provisions for assessing risks related to cybersecurity threats to critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) and biochemical weapon development. (10) "Frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10)" means a person who has trained, or initiated the training of, a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11), with respect to which the person has used, or intends to use, at least as much computing power to train the frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11). (11)(a) "Frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11)" means a foundation modelFoundation model"Foundation model" means an artificial intelligence model that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks.R.S. 51:3111.2(A)(8) that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillationKnowledge distillation"Knowledge distillation" means any supervised learning technique using a larger artificial intelligence model or its output to train a smaller model with similar or equivalent capabilities.R.S. 51:3111.2(A)(12) to a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillationKnowledge distillation"Knowledge distillation" means any supervised learning technique using a larger artificial intelligence model or its output to train a smaller model with similar or equivalent capabilities.R.S. 51:3111.2(A)(12) exceeds five million dollars. (b) The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation modelFoundation model"Foundation model" means an artificial intelligence model that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks.R.S. 51:3111.2(A)(8). (12) "Knowledge distillationKnowledge distillation"Knowledge distillation" means any supervised learning technique using a larger artificial intelligence model or its output to train a smaller model with similar or equivalent capabilities.R.S. 51:3111.2(A)(12)" means any supervised learning technique using a larger artificial intelligence modelArtificial intelligence model"Artificial intelligence model" means a machine-based system that receives inputs and generates outputs, including but not limited to predictions, content, recommendations, or decisions that can influence physical or virtual environments. "Artificial intelligence model" includes machine learning models, deep learning models, and other computational models designed to simulate aspects of human intelligence.R.S. 51:3111.2(A)(2) or its output to train a smaller model with similar or equivalent capabilities. (13) "Large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)" means a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) that together with its affiliatesAffiliate"Affiliate" means a person controlling, controlled by, or under common control with a specified person.R.S. 51:3111.2(A)(1) had more than five hundred million dollars in annual gross revenues in the preceding calendar year. (14) "Model weightModel weight"Model weight" means a numerical parameter in a frontier model that is adjusted through training and helps determine how inputs are transformed into outputs.R.S. 51:3111.2(A)(14)" means a numerical parameter in a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) that is adjusted through training and helps determine how inputs are transformed into outputs. (15) "PropertyProperty"Property" means corporeal or incorporeal property.R.S. 51:3111.2(A)(15)" means corporeal or incorporeal propertyProperty"Property" means corporeal or incorporeal property.R.S. 51:3111.2(A)(15).

A 1 A large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall write, implement, comply with, and clearly and conspicuously publish on its internet website a frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) that applies to the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)'s frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) and describes in detail how the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) approaches all of the following: (1) Incorporating national and international standards, and industry-consensus best practices into its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9), including standards related to cybersecurity and biosecurity risk management. (2) Defining and assessing the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)'s thresholds to identify and assess whether a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) has capabilities that may pose a catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3), including specific thresholds for cybersecurity capabilities that may be used to compromise critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) and capabilities, and that may assist in the development, production, or deployment of chemical or biological weapons. These thresholds may include multiple-tiered levels of concern. (3) Applying mitigations to address potential catastrophic risksCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) identified under Paragraph (2) of this Subsection, with specific mitigation strategies for cybersecurity risks to critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) and biochemical weapon risks. (4) Reviewing assessments and adequacy of mitigations as part of the decision to deployDeploy"Deploy" means to make a frontier model available to a third party, except for the primary purpose of developing or evaluating the frontier model.R.S. 51:3111.2(A)(7) a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) or use it extensively and internally. (5) Using third parties, including cybersecurity experts and biosecurity specialists, to assess catastrophic risksCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) and mitigation effectiveness. (6) Revisiting and updating the frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9), including any criteria that trigger updates and how the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) determines when its frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) are substantially modified enough to require disclosures pursuant to Subsection C of this Section. (7) Cybersecurity practices to secure unreleased model weightsModel weight"Model weight" means a numerical parameter in a frontier model that is adjusted through training and helps determine how inputs are transformed into outputs.R.S. 51:3111.2(A)(14) from unauthorized modification or transfer, including measures to prevent state-sponsored or other sophisticated threat actors from accessing model capabilities. (8) Identifying and responding to critical safety incidentsCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5), with specific protocols for incidents involving cyberattacks on critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) or potential biochemical threats. (9) Instituting internal governance practices to ensure implementation of these processes. (10) Assessing and managing catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) resulting from the internal use of its frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11), including risks resulting from a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) circumventing oversight mechanisms.

B 2 B.(1) A large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall review and update its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) at least once per year. (2) Within thirty days of making any material modification to its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9), a large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall clearly and conspicuously publish on its website the modified frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) and a justification for the modification.

C(1) 3 C.(1) Before, or concurrently with, deploying a new or substantially modified frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11), a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall clearly and conspicuously publish on its website a transparency report containing the frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11)'s release date, languages, output modalities, intended uses, and restrictions or conditions, and a mechanism to communicate with the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10).

C(2)–(3) 4 Before, or concurrently with, deploying a new or substantially modified frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11), a large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall include in the transparency report required by Paragraph (1) of this Subsection summaries of all of the following: (a) Assessments of catastrophic risksCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) from the frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) conducted pursuant to the frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9), including specific assessments of cybersecurity capabilities and biochemical risk. (b) The results of those assessments. (c) The extent to which third-party evaluators, including cybersecurity and biosecurity experts, were involved. (d) Other steps taken to fulfill the requirements of the frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) with respect to the frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11). (3) A frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) that publishes the information under this Subsection in a larger document, including a system card or model card, shall be in compliance with the applicable provision of this Section.

D 5 D. A large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall transmit to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) a summary of any assessment of catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) resulting from internal use of its frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) every three months or pursuant to another reasonable schedule specified by the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) and communicated in writing to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6).

E 6 E. A frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall not make a materially false or misleading statement about catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3), its management of catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3), or its implementation of, or compliance with, its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9), including but not limited to in a certification filed to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) pursuant to Subsection G of this Section, unless the statement was made in good faith and was reasonable under the circumstances.

F 7 F.(1) Beginning on July 1, 2028, and at least annually thereafter, a large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall retain a third-party auditor to produce a report signed and certified by the auditor that contains but is not limited to both of the following: (a) A detailed assessment of whether the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) has substantially complied with its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) during the prior year, including cybersecurity and biochemical risk management provisions. (b) Any material deviations between the requirements of the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)'s frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) and the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)'s actual practices during the prior year. (2) The third-party auditor shall not conduct an audit if it has a financial interest in the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) other than financial compensation for performing an audit. (3) Within thirty days of receiving an auditor's report required by Paragraph (1) of this Subsection, the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall publish on its website a high-level summary of the findings of the audit.

G 8 G. Within twelve months after publishing its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) in accordance with Subsection A of this Section, and annually thereafter, a large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall provide to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) a written certification disclosing either of the following: (1) That the developer has been in compliance with its frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) during the prior year. (2) Any material deviations from the developer's frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9) that occurred during the prior year, including a description of corrective actions taken or planned.

H 9 H.(1) When a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) publishes documents to comply with the provisions of this Section, the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) may make redactions to those documents as necessary to protect the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10)'s trade secrets or cybersecurity, public safety, or United States national security or to comply with any federal or state law. (2) If a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) redacts information in a document pursuant to this Subsection, the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall describe the character and justification of the redaction in any published version of the document to the extent permitted by the concerns that justify redaction and shall retain the unredacted information for five years.

A A. The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall establish a mechanism for frontier developersFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) or the public to report critical safety incidentsCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) that includes all of the following information: (1) The date of the critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5). (2) The reasons the incident qualifies as a critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5). (3) A description of the incident, including whether the incident involved cybersecurity threats to critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) or biochemical risks. (4) Whether the incident involved internal use of a frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11). (5) Whether the incident posed any risk to energy, health care, or port infrastructure in the state or in any other jurisdiction.

B B.(1) The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall establish a mechanism for large frontier developersLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) to confidentially submit summaries of any assessments of potential catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) from internal use of its frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11). (2) The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall take all precautions to limit access of any summaries submitted pursuant to Paragraph (1) of this Subsection to only personnel with a need to know and in order to prevent unauthorized access.

C 10 C.(1) Subject to the provisions of Paragraph (2) of this Subsection, a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall report any critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) pertaining to its frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) within fifteen days after discovering the critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5). (2) If a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) discovers that a critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) poses an imminent risk of death, serious physical injury, or an active cyberattack on critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4), the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall disclose that incident within twenty-four hours to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) and to any other appropriate authority with jurisdiction as required by law. (3) A frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) may file an amended report upon discovering additional information about the critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5).

D–E D. The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall review critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) reports submitted by frontier developersFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) and may review reports submitted by the public. E.(1) The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) may transmit reports of critical safety incidentsCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) and reports from covered employeesCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) to the legislature, the governor, the federal government, or appropriate state or federal agencies, including the Cybersecurity and Infrastructure Security Agency and the United States Department of Homeland Security. (2) Beginning July 1, 2028, and annually thereafter the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall: (a) Produce a report with anonymized and aggregated information about critical safety incidentsCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) reviewed since the preceding report, with specific sections addressing incidents related to cybersecurity and biochemical risks. (b) Produce a report with anonymized and aggregated information about covered employeesCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2)' reports reviewed since the preceding report. (3) A report transmitted pursuant to this Subsection shall not include information that may compromise the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10)'s trade secrets or cybersecurity, public safety, or national security, or violate any federal or state law. (4) The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall submit the reports to the president of the Senate, the speaker of the House of Representatives, and the governor.

F 10 F.(1) The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) may designate federal laws, regulations, or guidance that impose substantially equivalent or stricter standards or requirements for critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) reporting. (2) After a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) has declared to the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) its intent to comply with a designated federal law, regulation, or guidance, the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall constitute compliance with this Section to the extent that the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) complies with the designated federal law, regulation, or guidance until the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) revokes its intent or the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) revokes the relevant designation. (3) Failure to meet these standards or requirements constitutes a violation of this Part. (4) The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall revoke a designation if the requirements of Paragraph (1) of this Subsection are no longer met.

A–B A. The following records are confidential and shall not be subject to the provisions of the Public Records Law, R.S. 44:1 et seq.: (1) A report of assessments of catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) from internal use pursuant to R.S. 51:3111.2. (2) A certification submitted pursuant to R.S. 51:3111.2. (3) A report of a critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5) submitted pursuant to R.S. 51:3111.3. (4) A covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) report. B. The provisions of this Section shall terminate on July 1, 2031.

A–C A. On or before July 1, 2028, and annually thereafter, the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6), in consultation with the Governor's Office of Homeland Security and Emergency Preparedness, shall assess developments relevant to this Chapter and recommend whether to update the definitions of "frontier modelFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11)", "frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10)", and "large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)" to reflect technological developments and widely accepted standards. B. In making recommendations pursuant to this Section, the departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall consider: (1) Similar federal or international thresholds for the management of catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3), aligning with federal definitions to the extent consistent with this Chapter. (2) Evolving cybersecurity threat landscapes affecting critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4), particularly energy, health care, and port sectors. (3) Developments in biochemical weapon capabilities enabled or facilitated by artificial intelligence. (4) Stakeholder input, including input from critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4) operators in this state. (5) The length, complexity, and external verifiability of coverage determinations. C. The departmentDepartment"Department" means the Louisiana Department of Justice, office of the attorney general.R.S. 51:3111.2(A)(6) shall submit its findings and recommendations, together with specific proposals for legislation, to the president of the Senate of the Louisiana Legislature and the speaker of the House of Representatives of the Louisiana Legislature.

A–D A. A large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) that fails to publish or transmit a compliant document required to be published or transmitted under this Chapter, makes a statement in violation of R.S. 51:3111.2(E), fails to complete a third-party audit or publish a summary of the results of an audit as required by R.S. 51:3111.2(F), fails to complete a certification as required by R.S. 51:3111.2(G), fails to report an incident as required by R.S. 51:3111.3, or fails to comply with its own frontier AI frameworkFrontier AI framework"Frontier AI framework" means documented technical and organizational protocols to manage, assess, and mitigate catastrophic risks, with specific provisions for assessing risks related to cybersecurity threats to critical infrastructure and biochemical weapon development.R.S. 51:3111.2(A)(9), is subject to a civil penalty. B. The attorney general may bring an action against a violator to recover a penalty not to exceed one million dollars per violation for a first violation or ten million dollars for a second or subsequent violation of the same requirement. C. For purposes of bringing an action pursuant to this Section, a frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) or large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) that develops, deploysDeploy"Deploy" means to make a frontier model available to a third party, except for the primary purpose of developing or evaluating the frontier model.R.S. 51:3111.2(A)(7), or operates frontier modelsFrontier model"Frontier model" means a foundation model that either: (i) Was trained using a quantity of computing power greater than 10^26 integer or floating-point operations; or (ii) Was produced by applying knowledge distillation to a frontier model as defined in this Subparagraph (a) of this Paragraph, provided that the compute cost for applying knowledge distillation exceeds five million dollars. The quantity of computing power described in Subparagraph (a) of this Paragraph includes computing for the original training run and any subsequent fine-tuning, reinforcement learning, or other material modifications to a preceding foundation model.R.S. 51:3111.2(A)(11) in this state is both engaged in substantial and not isolated activities within this state, and is doing business in this state, shall be subject to the jurisdiction of the courts of this state. D. Loss of value of equity does not count as damage to or loss of propertyProperty"Property" means corporeal or incorporeal property.R.S. 51:3111.2(A)(15) for the purposes of this Chapter.

This Chapter preempts any rule, regulation, code, ordinance, or other law adopted by a parish, municipality, or other local governmental entity on or after July 1, 2027, specifically related to the regulation of frontier developersFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) with respect to their management of catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3).

A A. As used in this Section, the following terms have the meanings ascribed to them herein: (1) "Catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3)" has the same meaning as provided in R.S. 51:3111.2, except that the term applies to a foundation modelFoundation model"Foundation model" means an artificial intelligence model that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks.R.S. 51:3111.2(A)(8). (2) "Covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2)" means an employee responsible for assessing, managing, or addressing risk of critical safety incidentsCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5). (3) "Critical safety incidentCritical safety incident"Critical safety incident" means any of the following: (a) Unauthorized access to, modification of, or exfiltration of the model weights of a frontier model that results in death, bodily injury, or a cyberattack against critical infrastructure. (b) Harm resulting from the materialization of a catastrophic risk as defined in this Section. (c) Loss of control of a frontier model causing death, bodily injury, or damage to critical infrastructure. (d) A frontier model that provides specific and actionable technical assistance for the creation of a chemical or biological weapon to a person who would not otherwise have been able to obtain that assistance. (e) A frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.R.S. 51:3111.2(A)(5)" has the same meaning as provided in R.S. 51:3112, except that the term applies to a foundation modelFoundation model"Foundation model" means an artificial intelligence model that is trained on a broad data set, designed for generality of output, and adaptable to a wide range of distinctive tasks.R.S. 51:3111.2(A)(8).

B 11 B. Prohibited actions. A frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall not adopt, enforce, or enter into any policy or contract that prevents a covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) from, or retaliates against a covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) for, making a protected disclosure pursuant to this Section, or disclosing information to the attorney general, a federal authority, a supervisor, or another covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) who has authority to investigate, discover, or correct the reported issue, if the covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) has reasonable cause to believe that the information discloses that the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10)'s activities pose a specific and substantial danger to the public's health or safety resulting in a catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3), including risks to critical infrastructureCritical infrastructure"Critical infrastructure" means systems, assets, and networks, whether physical or virtual, that are so vital to the state or the United States, that the incapacity or destruction of these systems, assets, or networks, would have a debilitating impact on public health, public safety, economic security, or any combination thereof. "Critical infrastructure" includes, without limitation, energy production and refining facilities, healthcare delivery systems, shipping port operations and logistics systems, water treatment and distribution systems, electrical generation and transmission systems, natural gas pipeline systems, and petrochemical manufacturing facilities.R.S. 51:3111.2(A)(4), or that the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) has violated R.S. 51:3111.2 or R.S. 51:3111.3.

C 12 C. A frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall provide, to all covered employeesCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2), a clear notice of their rights pursuant to this Section by as follows: (1) Posting and displaying, at all times, within any workplace maintained by the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10), a notice of rights to all covered employeesCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2)', to ensure that any new or remote covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) receives equivalent notice. (2) Providing annual, written notice to each covered employee of their rights under this Section, to be acknowledged by each covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2).

D 13 D. Reporting process. (1) A large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) shall provide an anonymous, internal process for covered employeesCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) to disclose, in good faith, information to the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) which indicates specific and substantial danger to the public's health or safety resulting in a catastrophic riskCatastrophic risk"Catastrophic risk" means a foreseeable and material risk that a frontier developer's development, storage, use, or deployment of a frontier model will materially contribute to the death or serious injury of more than fifty people or more than one billion dollars in property damage or loss arising from a single incident involving a frontier model: (i) Providing expert-level assistance in the creation, synthesis, acquisition, weaponization, or release of a chemical or biological weapon, including but not limited to the identification of novel chemical compounds or biological agents with potential for weaponization, or the provision of detailed technical instructions for the production, stabilization, or dispersal of such agents. (ii) Engaging in or materially facilitating a cyberattack, with no meaningful human oversight, intervention, or supervision, against critical infrastructure systems, including but not limited to energy production and distribution systems, healthcare information technology systems, port logistics and navigation systems, water treatment facilities, or electrical grid systems, where cyberattacks result in or poses a substantial risk of physical harm, environmental damage, or significant disruption of essential services. (iii) Engaging in conduct with no meaningful human oversight, intervention, or supervision that, if the conduct had been committed by a human, would constitute a felony under state or federal law, including but not limited to offenses involving extortion, fraud, theft, or unauthorized access to computer systems. (iv) Evading the control of its frontier developer or user in a manner that demonstrates autonomous pursuit of objectives inconsistent with the developer's or user's instructions or safety protocols. "Catastrophic risk" does not include a foreseeable and material risk from any of the following: (i) Information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model. (ii) Lawful activity of the state or federal government. (iii) Harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.R.S. 51:3111.2(A)(3) or a violation of R.S. 51:3111.2 or R.S. 51:3111.3. The process shall include a monthly update, to the covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2), regarding the status of the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13)'s investigation of the disclosure and the actions being taken by the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) in response to the disclosure. (2) The disclosures and responses of the process pursuant to this Subsection shall be shared with the officers and directors of the large frontier developerLarge frontier developer"Large frontier developer" means a frontier developer that together with its affiliates had more than five hundred million dollars in annual gross revenues in the preceding calendar year.R.S. 51:3111.2(A)(13) on a quarterly basis, unless the officer or director is the subject of the disclosure.

E–F E. In a civil action pursuant to this Section, once it has been demonstrated by a preponderance of the evidence that a proscribed activity contributed to the alleged prohibited action by the covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2), the frontier developerFrontier developer"Frontier developer" means a person who has trained, or initiated the training of, a frontier model, with respect to which the person has used, or intends to use, at least as much computing power to train the frontier model.R.S. 51:3111.2(A)(10) shall prove, by clear and convincing evidence, that the action occurred for legitimate, independent reasons. F. A covered employeeCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2) aggrieved by a violation of this Section may bring a civil action to a court of competent jurisdiction for appropriate temporary, preliminary, or permanent injunctive relief. The court shall consider harm from the violation and any chilling effect on other covered employeesCovered employee"Covered employee" means an employee responsible for assessing, managing, or addressing risk of critical safety incidents.R.S. 51:3111.9(A)(2). The court may award reasonable attorney fees to the prevailing plaintiff.