AI Frontier Institute

In culpability, providing for liability for deployment of artificial intelligence system.

View primary source →

Pennsylvania House Bill 1533 (2025) has been referred to the House Judiciary Committee and has not been enacted. If enacted, it would add 18 Pa.C.S. § 306.1, making anyone who deploys an artificial intelligence system to perform an automated or autonomous function subject to criminal liability, civil liability, or both, for any negative outcome of the deployment. Section 306.1(a) lists covered harms: physical harm from AI-driven machinery, robotics or self-driving vehicles; economic misconduct such as price-fixing, bid-rigging, market allocation or fraud; unlawful data scraping, copyright infringement, IP theft and privacy violations; discriminatory, defamatory or harmful decisions from algorithmic bias; false or misleading AI-generated statements; and any other act that would be a legal violation or tort if committed by a person. Subsection (b) says automating an action does not remove a person's common-law duty of care. Subsection (c)(1) bars defenses that the AI acted autonomously, that outcomes were unintended or unforeseen, that a third party built or provided the system, or that it was marketed or certified as "safe" or "autonomous." Subsection (c)(2) allows two defenses: (i) reasonable ongoing oversight, safeguards and fail-safes with timely corrective action, or (ii) harm caused solely by unforeseeable, unauthorized outside interference despite reasonable security measures. Under (c)(3), failing to establish a defense means full liability. Subsection (d) defines "artificial intelligence system" (including generative AI) and "deployment" (which covers commercial, enterprise and individual use but excludes secured, purely experimental, nondeployed research models). Under Section 2, the act would take effect 60 days after enactment.

← Back to the tracker