EU Digital Law: Inferential Capability Does Not Define Legal Scope

A recent paper published on arXiv (2608.10601) posits that the legal boundaries of EU digital regulations are not defined by a technology's inferential abilities. Titled "Inferential Capability Does Not Determine Legal Scope," the study focuses on two significant EU digital regulations: the AI Act and the General Data Protection Regulation (GDPR). Article 3(1) of the AI Act uses inferential capability as a defining criterion, distinguishing regulated AI from standard software. Conversely, the GDPR does not specify inference but addresses it in a protective manner, linking legal ramifications to personal data processing and the effects of inferences on individuals, irrespective of the technology's classification as AI. The authors argue that these laws do not overlap, a discrepancy that becomes more pronounced with agentic architectures. Their main assertion is that inferential capability does not dictate legal scope, and its lack does not confer immunity. The proposed framework operates on two levels, with inference serving both constitutive and protective roles. The protective aspect involves three pathways, although the abstract does not elaborate on these. This paper contributes to the ongoing discussions surrounding AI regulation and data protection within the European Union.