Bill > S0358

This bill introduces a comprehensive legal framework for holding artificial intelligence (AI) developers strictly liable for injuries caused by their AI systems, specifically focusing on "covered models" and their derivatives that meet certain computational and cost thresholds. The legislation defines AI as a machine-based system that can infer and generate outputs influencing physical or virtual environments, and establishes a legal standard where developers can be held responsible for injuries caused by AI systems that would be considered negligent, intentional torts, or criminal if performed by a human. The bill creates a rebuttable presumption about the mental state of the AI system based on how a human would likely act in similar circumstances and provides two potential affirmative defenses for developers: demonstrating the AI met the standard of care applicable to humans performing the same function, or proving the injury resulted from a "capabilities failure" where the AI did not perform as intended but would not meet the legal standard of negligence. Key technical terms like "fine-tuning" and specific computational thresholds are precisely defined, and the legislation applies specifically to non-user injuries where the AI's conduct could be considered negligent or criminal. The bill aims to create a clear legal pathway for individuals seeking compensation for harm caused by advanced AI systems, effectively treating these systems as potential legal actors with defined standards of accountability.

Justice

https://status.rilegislature.gov/