The Effect of Deepfake Video on News Credibility and Corrective Influence of Cost-Based Knowledge about Deepfakes

Abstract

With rapid technical advancements, deepfakes—i.e., hyper-realistic fake videos using face swaps—have become more widespread and easier to create, challenging the old notion of “seeing is believing.” Despite raised concerns over potential impacts of deepfakes on people’s credibility toward audio-visual evidence in journalism, systematic investigation of the topic has been lacking. This study conducted an experiment (N = 230) that tested (1) how a news article using deepfake video (vs. real video) affects news credibility and viral behavioral intentions and (2) whether, based on signaling theory, obtaining knowledge about the low cost of producing deepfakes reduces the impact of deepfake news. Results show that people whose pre-existing attitudes toward controversial issues (abortion, marijuana legalization) are congruent with the advocated position of a news article are more likely to believe and be willing to share deepfake news as much as real video news. In addition, educating participants about the low cost of producing deepfakes was effective in reducing the credibility and viral behavioral intention of deepfake news for those who have congruent issue attitudes. This study provides evidence for differing levels of susceptibility for deepfake news and the importance of media literacy education regarding deepfakes that would prevent biased reasoning.

Keywords:

• Deepfakes
• fake news
• news credibility
• viral behavioral intention
• signaling theory
• media literacy education

Disclosure Statement

There is no competing interest with the research.

Notes

1 A priori power analysis using G*Power indicated that N = 158 was needed to detect medium effect (Cohen’s f = 0.25) using a one-way ANOVA comparing three news conditions (1 − β = .80; α = .05).

2 The region of significant moderation identified by Johnson-Neyman technique was from −2.78 to −0.21 values of mean-centered issue attitudes.

3 The region of significant moderation identified by Johnson-Neyman technique was from 0.54 to 3.42 values of mean-centered issue attitudes.

Additional information

Funding

This work was supported by University of Hawaii at Manoa and The University of Alabama.