<![CDATA[Advancements in Artificial Intelligence (AI) technology have given rise to deepfake technology capable of manipulating a person's image, voice, and identity with a high degree of realism. This phenomenon has spawned a new form of cybercrime—cyber defamation—that inflicts compounding harm on victims. This study aims to analyze the legal vacuum regarding the liability of AI developers in Indonesia and to formulate an ideal legal liability framework for developers for the future (ius constituendum). The study employs a normative legal research method, utilizing statutory, conceptual, and comparative approaches—drawing comparisons with the jurisdictions of the United States, Singapore, and the European Union. The findings indicate that current Indonesian positive law—specifically the ITE Law and the Criminal Code—focuses primarily on the content disseminator (end-user) and fails to address the legal obligations of developers regarding the misuse of dual-use technology. There are legal obstacles to applying the doctrine of complicity (deelneming) and difficulties in establishing a causal nexus between developer negligence and the harm suffered by the victim. Consequently, a legal reconstruction is required, incorporating the doctrine of strict liability for developers of high-risk AI systems, alongside legally binding obligations for "Safety by Design" and "Ethics by Design." Furthermore, a multi-layer liability model involving developers, platform providers, and end-users should be integrated to ensure proportional justice. The study recommends revising cyber legislation to mandate digital watermarking for synthetic media and harmonizing administrative and corporate criminal sanctions for AI developers who disregard safety standards during system design]]>
