Believing the unreal: A systematic review in mapping human responses to deepfakes through empirical evidence

Deepfakes use artificial intelligence to create manipulated media that look so realistic, they are increasingly shaping how individuals interact with information online. However, urgent concerns have been raised about their impact on public and media literacy. While technical advancements in detection are growing, the human factors influencing perceptions and behaviors, and the impact of deepfakes on society and culture remain underexplored. To address this gap, this systematic literature review synthesizes 66 empirical studies (2017–2025) across ten academic databases following PRISMA guidelines. The review categorizes research into four areas: user perception and behavior, societal impact, cultural context, and educational innovation. The results show that users often use heuristic cues, like facial artifacts or emotional resonance, to judge authenticity, but they overestimate their detection abilities. People usually believe and share manipulated media because their emotions, political perspectives, and social motivations influence their way of analytical reasoning. Cultural norms and media environments significantly mediate perception, with differences observed across regions like the U.S., China, and Singapore. While several educational strategies demonstrate potential in improving critical media literacy, their adoption remains limited. The review finds that there is a lack of long-term studies that include people from different cultures or use the same theoretical framework. It concludes by calling for theory-driven educational and policy frameworks that are more inclusive and flexible. This review guides researchers, educators, developers, and policymakers working to improve societal resilience against media manipulation so that they can effectively respond to deepfakes' evolving threats. • Analyzed 66 empirical studies on human perception and behavior toward deepfakes. • Identifies key users and social factors shaping detection and sharing. • Maps cultural differences in deepfake interpretation across regions. • Reviews the effectiveness of awareness and media literacy in enhancing detection interventions. • Proposes a factor-level framework to guide future research and educational strategies.