Human Attribution of Causality to AI Across Agency, Misuse, and Misalignment

New research examines how humans assign causal responsibility when AI systems are involved in harmful outcomes, finding that people attribute greater blame to AI when it has moderate to high autonomy, but still judge humans as more causal than AI when roles are reversed. The study provides insights for developing liability frameworks as AI incidents become more frequent and severe.

• →Humans attribute greater causal responsibility to AI when it has moderate or high agency in determining goals and means.
• →People consistently judge humans as more causal than AI even when both perform identical actions in reversed roles.
• →AI developers are perceived as highly responsible for harmful outcomes despite being temporally distant from incidents.
• →The agentic component of AI systems is judged as more causally responsible than passive language model components.
• →These findings will inform liability frameworks and policy debates around AI-caused harms as incidents increase.