Reconstructing Criminal Liability through the Actuant: Addressing Responsibility Gaps in Artificial Intelligence Systems

Abstract

Objective: To analyze the insufficiency of the subject object bipartite model in contemporary criminal law in the face of the technical autonomy of artificial intelligence systems and to propose the intermediate category of actuant as a dogmatic tool for reconstructing criteria of criminal attribution.

Method: This study adopts a qualitative theoretical and dogmatic approach with hermeneutic and comparative analysis of criminal law doctrine and philosophy of technology. Classical categories of the theory of crime, contributions from actor network theory, and normative experiences related to artificial intelligence were examined to identify structural gaps in criminal responsibility.

Results: The findings demonstrate that the traditional dichotomy between subject and object is unable to adequately explain situations in which intelligent systems perform criminally relevant actions without direct human intervention. The introduction of the category of actuant allows for the rearticulation of the theory of action and the criteria of objective and subjective attribution, preserving the principles of culpability and legality without assigning legal personality to machines. The proposed tripartite structure contributes to reducing responsibility gaps and restoring the traceability of criminal action in contexts of technological delegation.

Conclusion: It is concluded that the category of actuant constitutes an essential theoretical instrument for adapting criminal law dogmatics to the sociotechnical reality of the twenty-first century. Its incorporation enables overcoming the structural void of attribution in crimes mediated by artificial intelligence without undermining the guarantee-based foundations of criminal law.