Deepfake Videos: Perceived Realism, Confidence, and Detection Ability

Abstract Book of the 7th World Conference on Media and Communication

Year: 2026

[PDF]

Deepfake Videos: Perceived Realism, Confidence, and Detection Ability

Emel Dikbaş Torun, Aysel Curavcı

ABSTRACT:

Deepfake videos raise questions about whether users can reliably distinguish manipulated audiovisual content from authentic content and how confident they are in these judgments. This study examines the relationship between the perception of authenticity in deepfake videos, confidence in authenticity judgments, and actual detection performance.
The study was conducted with undergraduate and graduate students at Pamukkale University. A 3 × 2 mixed experimental design was used. Participants were assigned to control, awareness, and incentive groups, while video type (authentic vs. deepfake) constituted the within-subject factor. Thirty videos, selected based on the evaluations of three experts, were used in the experiment: 15 authentic and 15 deepfake. After each video, participants indicated whether they considered it authentic or deepfake and reported their confidence in this judgment. At the end of the experiment, they also estimated the number of videos they had identified correctly and described the cues they used when making their decisions. The findings showed that participants’ detection performance remained close to the 50% chance level. Despite this relatively low accuracy, confidence in their judgments remained high, indicating a clear discrepancy between actual performance and perceived ability. Age, gender, education level, and daily internet use were not significantly associated with detection performance. Participants also relied largely on visual and intuitive cues when deciding whether a video was authentic or manipulated. The findings suggest that familiarity with digital media does not necessarily translate into an ability to identify deepfake content. The gap between detection performance and confidence is particularly relevant to discussions of trust and judgment in digital media.

Keywords: Deepfake, Detection Ability, Confidence, Authenticity, Digital Media