Challenges in Regulating and Prosecuting AI Model Poisoning as Cybercrime

Abstract

This research examines the critical challenges confronting legal systems in regulating and prosecuting AI model poisoning as cybercrime. Through qualitative doctrinal analysis and comprehensive document review, the study evaluates international legal frameworks addressing AI poisoning, explores prosecution difficulties including proving intent and establishing liability, and assesses regulatory roles in preventing incidents. Findings reveal significant gaps in existing cybercrime statutes that fail to recognize AI poisoning as distinct offenses, creating uncertainty for law enforcement. The automated nature of machine learning obscures causation chains, making liability determinations nearly impossible under traditional legal principles. Cross-border enforcement fails because international agreements like the Budapest Convention lack specific provisions for AI attacks spanning multiple jurisdictions. Courts operate without precedents, forcing reliance on inadequate analogies to conventional cybercrimes. The research recommends enacting comprehensive legislation explicitly criminalizing AI poisoning, updating international treaties to facilitate cooperation, establishing mandatory security standards for high-risk systems, and developing specialized forensic capabilities within law enforcement agencies to address these emerging technological threats effectively.