”JAG SKITER VÄL EGENTLIGEN I HUR TROVÄRDIGT DET ÄR, OM DET BARA ÄR FÖR ATT FÅ NÅGON ATT GARVA”. En kvalitativ studie om ungas förhållningssätt till AI-genererat deepfake-innehåll på sociala medier

Abstract

AI-generated manipulated content is rapidly changing the nature of the modern media environment. With only a few simple actions, anyone can create and distribute manipulated videos online. As technological development is transforming social media platforms, the importance of adopting a critical approach to digital content also increases. Manipulated information also poses societal and democratic challenges by reducing public trust in media and institutions as well as undermining people’s ability to remain well-informed. Although research on media and information literacy, AI and disinformation is extensive, the rapid change of technological development creates a need for continuous research. Emerging digital phenomena, such as deepfakes, should therefore be studied in relation to evolving patterns and developments, where human perspectives and interpretations remain central. The purpose of this study is to contribute to a deeper understanding on how young people assess the credibility of AI-generated deepfake content on social media platforms. This is examined by analysing the strategies used by young people when evaluating credibility, as well as the factors that determine when and why such strategies are activated. Through this, the study aims to provide greater insight into how young people reason about manipulated content encountered on social media platforms. By examining how adolescents themselves approach on deepfakes and their credibility, the study contributes to a broader understanding of critical media literacy practice that may help prevent the spread of disinformation. The study is based on a qualitative research design using focus group interviews with 23 upper secondary school students aged 16 to 19. Deepfake videos were used as video-material to encourage discussion about how participants perceive and assess the credibility of AI-generated content. The focus group method enabled a more in-depth understanding of how young people reason about manipulated media. The findings show that young people use several different strategies when assessing the credibility of deepfake material. In most cases, the findings show that credibility assessment should be understood as a situation-based process in which strategies may overlap. An important contribution to the study is the strategy we identified as the “Barbie-effect”. This refers to situations where content is perceived as less credible when it appears to be perfect or artificial. This may involve the absence of human skin texture or a visual quality that is perceived as unnaturally flawless. The results show that it is not only imperfections that indicate that content is manipulated; excessive perfection can also reduce the perception of authenticity. Other identified strategies for credibility assessment include auditory deviations, graphical deviations, and deviations in the content itself. Another part of the findings is that respondents use social signals from platforms in the credibility assessment. Comment sections and likes are used as indicators of how the content should be interpreted and evaluated. The results also show that credibility assessment is highly situational. Young people tend to activate more critical and strategic credibility assessments when the content is perceived as important, harmful, realistic or socially significant, for example in relation to politics, war or content that may influence people’s opinions. However, these strategies may also be absent or less developed when the content is perceived as humorous, entertaining, or irreverent. In such situations, the respondents often did not consider it necessary to examine whether the content was authentic or manipulated. The findings therefore that credibility assessment is not only about identifying technical deviations but are also shaped by the topic of the content, its perceived significance and previous knowledge. The findings highlight young people's awareness of AI-generated manipulated content, while also pointing to the broader importance of media and information literacy.