From creative output to creative governance: Legitimizing AI-generated creativity under deepfake uncertainty

Generative artificial intelligence challenges established foundations of innovation legitimacy by decoupling output quality from human effort, intention, and responsibility. When creative outputs no longer provide reliable signals of human agency, legitimacy judgments must be reconstructed through alternative interpretive cues. This study reconceptualizes innovation legitimacy as an inferential process formed under conditions of technological opacity and examines how governance practices shape legitimacy judgments in AI-mediated creativity. Drawing on a survey-based experiment (N = 240), we investigate how configurations of AI involvement, transparency timing, and deepfake-related manipulation cues influence perceived deception, perceived human effort, and perceived authenticity. Legitimacy judgments, operationalized as inferential assessments of deception, effort, and authenticity, are distinguished from their downstream attitudinal and behavioral consequences. The findings show that legitimacy in AI-mediated creativity shifts from output-based evaluation toward governance-dependent sensemaking. Transparency operates as an interpretive mechanism whose effects depend critically on timing, while deepfake cues systematically heighten moral uncertainty and undermine legitimacy. By specifying the microfoundations through which legitimacy breaks down and is partially reconstructed under AI-induced opacity, the study advances innovation theory by reframing legitimacy as a governance-dependent inferential achievement. This reframing differs from existing interpretive and institutional accounts: governance is conceptualized not as a mechanism that generates legitimacy directly through compliance or ceremonial conformity, but as an interpretive infrastructure that shapes the conditions under which audiences can form legitimacy judgments when conventional output-based cues are absent. The findings contribute to debates on AI governance, innovation diffusion, and responsible innovation in contexts characterized by ambiguous agency and uncertainty about algorithmic production processes.