AGAINST THE BLANKET APPLICATION OF THE HEIGHTENED DANGER REGIME TO ARTIFICIAL INTELLIGENCE SYSTEMS: A COMPARATIVE ARGUMENT FOR USE BASED DIFFERENTIATION

Abstract

Conflict of laws rules identify the applicable law but leave untouched the substantive basis of liability. In the doctrine of the Commonwealth of Independent States that basis is increasingly sought in the construct of the source of heightened danger, and a growing body of scholarship proposes to read artificial intelligence systems into it. This paper argues that the proposal is unfounded as a general rule. Two objections are advanced. The first is conceptual: the institution was formed for activities whose danger is physical, whereas most harm produced by artificial intelligence is proprietary or informational, so that extension destroys the internal logic of the institution. The second concerns the cross-border dimension: the construct has no direct equivalent in common law systems, and its extension distances national law from international practice at the stage of recognition and enforcement. Comparative examination of the European Union, China and the United States shows a structural convergence. None of the three has adopted a technology wide no-fault regime; in each, liability without fault is channelled through defectiveness or through physical danger. The paper proposes use based differentiation as the solution best suited to Uzbekistan: the heightened danger regime should apply only where the system governs a physical process endangering life or health, while all remaining cases should be resolved under the rules on defects in goods, works and services, supported by a documentary disclosure duty.