State-Level AI Agent Liability Frameworks Remain Nascent

Legal experts are examining the liability implications of Artificial Intelligence (AI) agent behaviour. Delaware proposes legislation to define AI, establish an AI advisory council, and develop ethical guidelines. These efforts address the complexity of assigning responsibility when AI systems act autonomously.

Traditional legal frameworks typically assign liability to human operators or manufacturers. AI agents, however, can initiate actions without direct human intervention. This capability challenges existing legal definitions of agency and intent. Questions arise over whether an AI’s developer, deployer, or user should bear the legal burden for unintended outcomes.

The University of Delaware established an AI Center in October. This centre aims to foster interdisciplinary research on AI's societal impact. Its focus includes legal, ethical, and policy considerations. The centre acknowledges the lack of established legal precedents for AI-driven incidents. This gap necessitates new approaches to regulation and jurisprudence.

Proposed Delaware legislation seeks to address some of these ambiguities. It defines AI as a ‘machine-based system’ that can ‘make predictions, recommendations, or decisions’. The bill also mandates risk assessments for certain AI deployments. However, it does not explicitly create new liability standards for AI agents. Scholars suggest a comprehensive approach must consider the full AI lifecycle, from design to deployment.

The increasing sophistication of AI agents complicates accountability. Their ability to adapt and learn may lead to actions unforeseen by their creators. Legal scholars are debating whether current product liability, negligence, or tort laws adequately cover these scenarios. Should states introduce specific AI liability statutes, or adapt existing legal principles?