Deepfake Manipulation on Social Media: A Case Study of Fraudulent Activities in Türkiye

The term "deepfake," derived from the words "deep" and "fake," refers to highly realistic and deceptive content designed to manipulate perception. Deepfake content based on artificial intelligence technology poses a threat when integrated with the possibilities of digital culture, and illegal activities are carried out by changing the faces and speeches of famous people. The study was conducted to examine social media-mediated deepfake fraud in Turkey. Facebook and Instagram platforms were scanned with four different accounts for a period of sixteen months and fifty-six deepfake videos were obtained. It was found that the videos included important political figures, scientists, artists, journalists-anchors, business people and all of them were fraudulent. It was observed that Lip-Sync (m = 0.75) was the primary model in the production of the videos, followed by Text-to-speech (m = 0.18), Lip-Sync & Text-to-speech (m = 0.5). The group with the most deepfake videos was business people (m = 0.35), followed by political figures (m = 0.33), scientists (m = 0.14), journalists-speakers (m = 0.12), artists (m = 0.03). The deepfake videos of Ali Koç, Ekrem İmamoğlu, Emine Erdoğan, Hakan Fidan and Murat Ülker, which were selected based on Lip-Sync performance, were analysed with Deepware and the detection architecture was found to be insufficient. In the research, which uses thematic analysis method and quantitative data in a mixed format, frequency distributions were created with IBM SPSS Statistics 22.0 programme. In the light of the findings, suggestions for the prevention of victimisation are presented.