Date of Award

August 2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Communication

First Advisor

Xudong Yu

Abstract

The proliferation of AI-generated media, particularly deepfakes, has generated growing concerns regarding misinformation, media authenticity, and public trust in digital communication. This study investigates how education about deepfakes influences the ability to detect deepfake and genuine videos as well as trust in media institutions. Drawing on Inoculation Theory, media literacy research, and sociological perspectives on misinformation and trust, this research examines whether educational materials about deepfakes enhance detection abilities while simultaneously reducing trust in media. A between-subjects experimental design was employed, with 220 undergraduate participants randomly assigned to either an educational treatment condition or a control condition. Pretest measures assessed media trust, perceived ability to detect misinformation, and verification behaviors. The treatment group viewed an educational video detailing the creation of deepfakes, their societal implications, and strategies for identifying deepfakes, while the control group viewed an unrelated video. Subsequently, participants completed a media identification task involving both authentic and AI-generated deepfake videos, followed by posttest measures assessing trust in media, verification intentions, concern about deepfakes, and intended media consumption.

The results showed that participants who received the intervention exhibited a decrease in their ability to detect authentic media, becoming more likely to misidentify genuine videos as deepfakes. ​However, the educational intervention did not noticeably reduce trust in media as a whole. Exploratory analyses showed no statistically significant effects on verification intentions, fact-checking behaviors, concern about deepfakes, or intended media consumption.

These findings contribute to a broader understanding of how authenticity and trust operate in the context of AI-generated media. By distinguishing between the ability to accurately detect genuine media content and trust in media institutions, this study provides insight into audience responses to deepfake education and identifies an important area for future sociological and communication research.

Download

Share

COinS