WHAT THIS BILL REGULATES · 1 REQUIREMENT TYPE
How Is This Bill Enforced
Verbatim statutory text on the left; plain-language analysis and a per-section checklist on the right. Numbered markers cross-link to the matching checklist row.
(a) 1 Liability.--A person that engages in the deployment of an artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d) to perform an automated or autonomous function shall be subject to criminal or civil liability, or both, for any negative outcome arising from the deploymentDeployment"Deployment." As follows: (1) The use of an artificial intelligence system in a manner that has the potential to affect external persons, systems or legal interests. (2) The term includes commercial implementation, enterprise use, individual use or the use of autonomous systems affecting third parties. (3) The term does not include the use of purely experimental, nondeployed research models if the models are reasonably secured to prevent unauthorized access and deployment.18 Pa.C.S. § 306.1(d), including: (1) Physical harm or endangerment caused by artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d)-driven machinery, robotics, self-driving vehicles or autonomous systems. (2) Economic misconduct, including price-fixing, bid-rigging, market allocation or fraud conducted through artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d)-driven decision making. (3) Unlawful data scraping, copyright infringement, intellectual property theft or violations of privacy and data protection laws committed through automated or autonomous systems. (4) Discriminatory, defamatory or otherwise harmful decision making arising from algorithmic bias or automated processing. (5) False, deceptive or misleading statements generated or distributed through artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d) models under the control of a person. (6) Any other act that, if committed by a person, would constitute a violation of law or amount to tortious conduct.
Subsection (a) establishes the bill's core liability rule: any person who deploys an AI system to perform an automated or autonomous function is subject to criminal or civil liability for any negative outcome arising from that deployment. The provision enumerates six non-exclusive categories of covered harms — physical harm from AI-driven machinery, economic misconduct through AI-driven decision-making, unlawful data scraping and IP violations, discriminatory or defamatory algorithmic outputs, false or deceptive AI-generated statements, and a catch-all for any act that would constitute a legal violation or tort if committed by a person. This is not an affirmative compliance obligation but a liability-assignment rule that ensures deployers cannot escape responsibility by interposing AI between themselves and the harmful outcome.
(b) 2 Automation.--Automating an action does not remove a person's general duty of care relating to the action as established under common law.
Subsection (b) codifies the principle that automating an action through AI does not remove a person's general common-law duty of care. This is a clarifying provision that prevents deployers from arguing that delegation to an AI system eliminated their duty of care — it reinforces that the duty attaches to the person who initiates the automated action, not the mechanism by which the action is carried out.
(c)(1) Except as provided in paragraph (2), a person may not avoid liability under this section by asserting that: (i) The artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d) acted autonomously or without explicit human instruction after the person engaged in the deployment of the artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d). (ii) The outcomes of the artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d)'s decision making were unintended, unforeseen or the result of machine-learning adaptation. (iii) The artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d) was trained, developed or provided by a third party and the person merely initiated the deployment of the artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d). (iv) The artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d) was marketed, certified or believed to be "safe," "self-regulating" or "autonomous."
(c)(2) 3 A person may assert a defense to liability under paragraph (1) by demonstrating that: (i) the person implemented reasonable, ongoing oversight, safeguards and fail-safe mechanisms designed to prevent unlawful, negligent or harmful conduct and took timely corrective action once the person became aware of a risk or failure relating to the artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d); or (ii) the artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d)'s harmful act or omission resulted solely from an unforeseeable and unauthorized interference by an external actor and the person had previously implemented reasonable security measures designed to prevent the unauthorized use.
(c)(3) The failure to establish a defense under paragraph (2) shall result in full liability for the deploymentDeployment"Deployment." As follows: (1) The use of an artificial intelligence system in a manner that has the potential to affect external persons, systems or legal interests. (2) The term includes commercial implementation, enterprise use, individual use or the use of autonomous systems affecting third parties. (3) The term does not include the use of purely experimental, nondeployed research models if the models are reasonably secured to prevent unauthorized access and deployment.18 Pa.C.S. § 306.1(d) as specified in this section.
Subsection (c) addresses the defense structure for AI deployment liability. Paragraph (1) eliminates four common defenses: that the AI acted autonomously, that outcomes were unintended, that a third party developed the AI, or that the AI was marketed as safe or autonomous. Paragraph (2) provides two affirmative defenses: demonstrating reasonable ongoing oversight with timely corrective action, or showing that the harm resulted solely from unforeseeable unauthorized external interference where reasonable security measures were in place. Paragraph (3) specifies that failure to establish either defense results in full liability.
The practical compliance implication is significant: to preserve the affirmative defense, deployers must implement and maintain reasonable, ongoing oversight, safeguards, and fail-safe mechanisms and take timely corrective action upon awareness of risk or failure. This is the closest the bill comes to imposing an affirmative compliance duty — deployers who wish to have any defense available must build and maintain an oversight and safety program.
(d) Definitions.--As used in this section, the following words and phrases shall have the meanings given to them in this subsection unless the context clearly indicates otherwise: "Artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d)." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed. "DeploymentDeployment"Deployment." As follows: (1) The use of an artificial intelligence system in a manner that has the potential to affect external persons, systems or legal interests. (2) The term includes commercial implementation, enterprise use, individual use or the use of autonomous systems affecting third parties. (3) The term does not include the use of purely experimental, nondeployed research models if the models are reasonably secured to prevent unauthorized access and deployment.18 Pa.C.S. § 306.1(d)." As follows: (1) The use of an artificial intelligence systemArtificial intelligence system"Artificial intelligence system." As follows: (1) A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments, including the ability to: (i) Perceive real and virtual environments. (ii) Abstract perceptions made under this paragraph into models through analysis in an automated manner. (iii) Use model inference to formulate options for information or action based on outcomes under subparagraphs (i) and (ii). (2) The term includes generative artificial intelligence and any substantially similar technology yet to be developed.18 Pa.C.S. § 306.1(d) in a manner that has the potential to affect external persons, systems or legal interests. (2) The term includes commercial implementation, enterprise use, individual use or the use of autonomous systems affecting third parties. (3) The term does not include the use of purely experimental, nondeployed research models if the models are reasonably secured to prevent unauthorized access and deploymentDeployment"Deployment." As follows: (1) The use of an artificial intelligence system in a manner that has the potential to affect external persons, systems or legal interests. (2) The term includes commercial implementation, enterprise use, individual use or the use of autonomous systems affecting third parties. (3) The term does not include the use of purely experimental, nondeployed research models if the models are reasonably secured to prevent unauthorized access and deployment.18 Pa.C.S. § 306.1(d).
Subsection (d) defines the bill's two key terms. Artificial intelligence system is defined broadly as any machine-based system that makes predictions, recommendations, or decisions influencing real or virtual environments, with capabilities including perception, automated abstraction into models, and model inference. The definition expressly includes generative AI and any substantially similar future technology. Deployment is defined as any use of an AI system that has the potential to affect external persons, systems, or legal interests, covering commercial, enterprise, and individual use as well as autonomous systems affecting third parties. A narrow carve-out excludes purely experimental, non-deployed research models that are reasonably secured against unauthorized access.
This act shall take effect in 60 days.
Section 2 provides that the act takes effect 60 days after enactment. No specific calendar date is stated because the bill has not been enacted.