Private Law Theory

ABSTRACT
When an autonomous robotic arm in a German automotive plant misidentified a component orientation and caused a collision that injured a line operator in 2024, six parties claimed they were not liable: the robot OEM (the arm performed within mechanical specifications), the AI vendor (the vision model met contractual accuracy thresholds), the system integrator (the deployment followed the vendor’s configuration guide), the plant operator (the safety protocols were those the integrator recommended), the component supplier (the part was within tolerance), and the insurance underwriter (the policy excluded ‘autonomous decision-making errors’). The injured worker’s claim took 14 months to resolve. The liability vacuum – the gap between existing legal frameworks and the distributed, probabilistic, and opaque nature of AI-driven decisions – is not a hypothetical future problem. It is a present operational reality.

This paper examines the emerging legal and contractual frameworks for distributing liability across the autonomous industrial systems value chain: OEMs who manufacture physical systems, system integrators who deploy and configure them, AI vendors who provide the intelligence layer, and operators who use them in production. Drawing on product liability law, contract theory, and insurance economics, the paper maps the current liability landscape across four jurisdictions (United States, European Union, India, and China) and identifies critical gaps in each. The paper proposes the Risk-Proportional Liability Framework (RPLF), which allocates liability based on three principles: control (who had the ability to prevent the harm), foreseeability (who could have anticipated the failure mode), and information asymmetry (who possessed knowledge about system limitations that others did not). The framework is designed to be implementable through commercial contracts, insurable through existing risk transfer mechanisms, and compatible with emerging regulatory requirements across jurisdictions.

Shaik, Ali Sadhik, Liability Allocation in Autonomous Industrial Systems: Who Pays when the AI is Wrong? (May 1, 2026).