[Legal responsibility in AI-assisted radiological reporting]

Artificial intelligence (AI) is increasingly used in radiological diagnostics, particularly for screening, detection, and prioritization of imaging exams. While these systems promise efficiency gains and improved diagnostic accuracy, their clinical integration raises important legal questions concerning responsibility and liability in cases of diagnostic error.

This article analyzes the legal framework governing AI in radiology and examines the allocation of liability between physicians, health care institutions, and manufacturers within existing medical liability structures.

A doctrinal legal analysis was conducted based on European and German legal sources, including the AI Act and the Medical Device Regulation, as well as provisions of German medical liability law.

AI systems in radiology are legally classified as medical tools rather than independent actors. Consequently, they cannot bear legal responsibility themselves. Liability primarily remains with the physician, who must critically review and validate AI-generated outputs within the diagnostic process. Manufacturers may be liable under product liability and regulatory compliance frameworks. The interaction between technological complexity and existing evidentiary standards creates increasing tension within traditional liability doctrines. Despite technological progress, the final diagnostic decision and legal responsibility remain with the physician. AI currently functions as a decision-support tool rather than a substitute for medical judgment. However, the increasing relevance of autonomous systems suggests that a distinct legal framework for AI in medicine may emerge in the future.