Artificial intelligence: defenses.

Assembly Member Krell's artificial intelligence liability measure establishes clear boundaries for civil cases involving AI systems by preventing defendants from deflecting responsibility onto autonomous AI actions. The legislation defines artificial intelligence as engineered or machine-based systems with varying levels of autonomy that process inputs to generate outputs affecting physical or virtual environments.

Under the new provisions, defendants who developed, modified, or used AI systems cannot claim in civil proceedings that autonomous AI behavior was solely responsible for harming a plaintiff. This builds upon existing California requirements for AI developers to publicly document their training data for systems released since January 2022. The measure maintains the fundamental principle that parties are responsible for injuries caused by their actions or negligence, whether through direct conduct or through the deployment of AI technologies.