<![CDATA[The development of artificial intelligence has transformed criminal evidence, particularly when electronic evidence is not merely falsified but adversarially manipulated to deceive AI-based analytical systems. This study aims to reconstruct the concept of obstruction of justice concerning adversarial manipulation of electronic evidence within Indonesia’s Criminal Code and Criminal Procedure Code. It employs normative legal research using statutory, conceptual, case, and comparative approaches. The findings show that adversarial manipulation differs from conventional digital falsification because attacks may target an AI system’s inference process without significantly altering the visible appearance of evidence. Normatively, Article 278 of the Criminal Code has the strongest connection to direct manipulation of evidence, while Articles 281 and 282 may only apply when their specific elements are satisfied. In procedural law, electronic evidence authentication should be supplemented by assessment of analytical reliability, including model transparency, error rates, reproducibility, and adversarial robustness. This study proposes the Adversarial Evidence Interference Test, consisting of judicial-process nexus, adversarial act, knowledge and purpose, material interference capability, and forensic causal traceability. This reconstruction enables criminal law to adapt to AI developments without expanding criminalization by analogy and strengthens protection for the integrity of criminal evidentiary processes. It promotes accountable, transparent, and reviewable digital adjudication. Keywords: Obstruction of justice; Adversarial manipulation; Electronic evidence; Artificial intelligence; Criminal evidence.]]>
