Visual Realism in AI-Generated Instagram Content: The Role of Perceived Manipulation in Brand Trust

Sammanfattning

The rapid adoption of generative artificial intelligence in digital marketing has enabled brands to produce highly realistic visual content at scale, yet the psychological mechanisms through which such imagery influences consumer brand trust remain underexplored. This research examines how visual realism in AI-generated Instagram advertising influences consumer brand trust, with perceived manipulation tested as the mediator. A quantitative, cross-sectional survey included 308 active Instagram users based in Sweden. The respondents saw two Instagram advertising visuals from an energy drink brand: one human-created and one AI-generated, before completing tested measures of visual realism, perceived manipulation, and brand trust. The AI-generated stimulus employed in this study was stylized rather than photorealistic. Visual realism was therefore measured as a respondent-level perceptual variable, not as an objective property of the stimulus itself. Mediation analysis was conducted using Hayes' (2018) PROCESS macro (Model 4) with 5,000 bootstrapped resamples. Contrary to the expected hypothesis (H1), higher perceived visual realism was associated with lower manipulation (B = −0.094, p = .027). However, (H2) when consumers perceived manipulation, brand trust declined accordingly (B = −0.114, p = .011). The proposed mediation pathway (H3) was not statistically significant (estimate = 0.011, 95% CI [−0.001, 0.032]), though visual realism predicted brand trust directly and positively (B = 0.081, p = .015). This indicates a more conditional relationship between AI-generated visual realism and brand trust than suggested by existing literature (Rapp et al., 2025; Liu & Chen, 2026; Park et al., 2024). One reason could be the visual aesthetic features of the AI-generated stimulus used in this research. Because the stimulus was clearly stylized rather than photorealistic, consumers likely read it as a creative aesthetic brand decision rather than as an attempt to deceive, explaining in part the lower levels of perceived manipulation. Results indicate that the stylistic presentation of AI-generated content, rather than AI authorship alone, should be more explicitly considered in theoretical models of consumer responses to AI-mediated advertising. The study thereby contributes empirical nuance to the growing literature on AI-generated visual content, persuasion knowledge, and consumer brand trust.