Combating the Spread of Deepfake Videos on Social Media with Labels
CSEF · 2023 Behavioral & Social Sciences Honorable_mention Award
Overview
Deepfake videos are the latest frontier on misinformation. One approach to combating these is to label fake videos with warnings. It is, however, currently not possible to detect all fake videos. This project assessed how three video labeling methods affect the perception of the videos. A group of 167 survey volunteers was recruited to rate 24 videos as real or fake, 12 of which were fake face-swap videos. Each volunteer was randomly assigned to one of four surveys. Survey 1 was the control with no video labels. Survey 2 had fake warning labels on half the fake videos. Survey 3 had fake warning labels on half of the fake videos and verified labels on half of the real videos. Survey 4 had the same labels as survey 3 but also added unverified labels on all the remaining unverified videos. In control survey 1, a staggering 50% of volunteers rated fake videos as real. This perception inaccuracy was not highly correlated with age, gender, education level, or social media familiarity. The most accurate perception results were obtained with survey 4’s labeling, where the believability of fake videos with fake labels and unverified labels were on average reduced by 19% and 6% respectively, but the believability of real videos with verified labels and unverified labels were on average increased by 8% and 10% respectively. These results suggest that social media platforms can slow the spread of fake video misinformation by labeling checked fake videos, checked real videos, and unchecked videos.
Source coverage
This record comes from a published award list, not a complete project archive. Its abstract comes from CSEF's public project showcase as archived by the Internet Archive before judging (https://web.archive.org/web/20230401224130/https://ca-csef.zfairs.com/showcase/ShowcaseInfo?f=838e60b7-ea75-46e8-865c-fde4864244b3); the version presented may differ.
Awards (1)
- Category Award: HM
Competition history
- CSEF 2023
Resources
Related projects
ISEF · 2025
A Novel Audio-Video Multimodal Deep Learning Model for Improved Deepfake Detection To Combat Disinformation
CSEF · 2018
Can a Preventative Social Media User Interface Break "Fake News"?
ISEF · 2022
Deepfake Detection Using Deep Learning
ISEF · 2018
Can a Preventative Social Media UI Break "Fake News?"
ISEF · 2024
Real or Fake? Using Artificial Intelligence to Detect Filtered Images
ISEF · 2021
Demistifying 'Fake News': Evaluating Media-Borne Misinformation Through the Novel Application of AI Powered Sentiment Analysis
ISEF · 2023
"The Truth Will Come Out": Defending Against the Viral Spread of Misinformation by Characterizing User Responses to Fake News
ISEF · 2022
Bringing Truth Into Fake News: The Diffusion of Fact-Checked Fake Stories on Twitter
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: California Science & Engineering Fair public projects