<p>Deep synthesis technology (DST), exemplified by deepfakes, raises significant ethical and legal concerns, yet users continue to actively engage with such applications even when these concerns are salient. This behavioral pattern presents a theoretical paradox: ethical risk does not uniformly suppress adoption behavior across contexts. Conventional technology adoption models such as UTAUT lack ethical constructs and may therefore prove insufficient for explaining adoption in morally contested technological contexts such as DST. Moreover, prior research has predominantly treated DST as a homogeneous category, overlooking variation across application types and the heterogeneous role of ethical factors in shaping adoption intentions. This study addresses these gaps by examining users’ adoption intentions across heterogeneous DST applications and uncovering underlying mechanisms. Integrating the unified theory of acceptance and use of technology (UTAUT) with Hunt-Vitell ethics theory, we conducted a scenario-based survey featuring utilitarian and hedonic DST applications, encompassing function-oriented contexts (e.g., public interest communication and organizational management) and entertainment-oriented contexts (e.g., celebrity impersonation and virtual performance). A multi-method strategy combining comparative analysis and configurational analysis assessed adoption differences and identified equifinal pathways. Results reveal significantly stronger adoption intentions toward utilitarian than hedonic DST. The two adoption contexts exhibit fundamentally distinct configurational logics. Utilitarian DST adoption follows a rational-instrumental logic, with descriptive norms serving as the core condition across all configurations. Performance expectancy, effort expectancy, and technology self-efficacy also emerge as frequently recurring core conditions across multiple pathways. Hedonic DST adoption, by contrast, is governed by an ethico-normative logic, with ethical acceptability serving as the core condition across all configurations. Injunctive norms, hedonic performance expectancy, and the absence of deepfake concern appear as core conditions across several pathways. These findings demonstrate that ethical risk does not uniformly suppress adoption behavior; rather, its causal role varies systematically across application contexts. This study extends UTAUT into ethically sensitive technological domains and offers differentiated, evidence-based guidance for DST governance and context-sensitive regulatory design.</p>