Acting with AI: An Interaction-Based Framework for Agentic Tort Liability
A new academic paper proposes an interaction-based legal framework for assigning liability when autonomous AI systems cause harm. The framework draws on theories of human planning and concerted action to categorize AI interactions into autonomous drift, pure tool use, or collaborative planning. It suggests using interaction logs as evidence to determine where responsibility should lie, proposing a 'Reasonable Agent' standard focused on verification, transparency, and logging. AI
IMPACT Proposes a novel legal framework for AI liability, potentially influencing future regulations and court decisions regarding AI-caused harm.