Navigating the Transparency Dilemma under the EU AI Act - How AI Disclosure Phrasing can Preserve Consumer Trust Through Perceived Effort

Sammanfattning

Purpose: This study investigates whether a novel 'Human-made, AI-enhanced' label preserves consumer trust in generative AI advertising under EU AI Act transparency requirements. By comparing this hybrid phrasing to standard disclosures, the research examines how signaling human effort preserves consumer trust. Theoretical Perspective: This study is grounded in effort heuristic theory, examining how varying degrees of signalled human involvement in AI-disclosure labels shape consumer trust through the cognitive mechanism of perceived effort. Method: A web-based experimental survey was conducted with 232 participants, recruited through convenience sampling. The study utilized an image-based advertisement to compare the effects of labels 'AI-generated', 'Human-made, AI-enhanced', and a no-label condition on dimensions of consumer trust and perceived effort. Results: The results demonstrate that the 'Human-made, AI-enhanced' label has higher levels of trust (Reliability and Usefulness) compared to standard 'AI-generated' disclosures. This relationship is mediated by perceived effort. Conclusion: Mandatory transparency is not an inherent brand liability. While standard disclosures trigger an effort penalty, collaborative phrasing resolves this psychological tension by highlighting human involvement. Theoretical Contributions: This work extends the effort heuristic to generative AI, demonstrating how the transparency dilemma is malleable rather than absolute when human agency is prefixed. Managerial Implications: Marketing practitioners should implement collaborative labeling like ‘Human-made, AI-enhanced’ to capitalize on generative AI cost-efficiencies and ensure advertisement trust