AN ACT RELATING TO COURTS AND CIVIL PROCEDURE -- PROCEDURE GENERALLY -- CAUSES OF ACTION

SB 358 creates a new legal right for people injured by artificial intelligence systems developed by companies that trained or fine-tuned very high-cost AI models (costing over $100 million for initial training or $10 million for significant modifications). It holds developers strictly liable for injuries to non-users (like bystanders) caused by AI systems that behave in ways a human would consider negligent or intentionally harmful, without requiring proof of developer negligence. The bill excludes defamation claims and establishes a rebuttable presumption that AI systems meet the mental state required for certain torts if a human in similar circumstances would have acted that way. Developers can avoid liability by proving the AI met human standards of care or the harm resulted from a "capabilities failure" (where the AI underperformed but didn't act negligently).