Examining the Implications of Deepfakes for Election Integrity

View on arXiv ← Back to list

Authors: Hriday Ranka, Mokshit Surana, Neel Kothari, Veer Pariawala, Pratyay Banerjee, Aditya Surve, Sainath Reddy Sankepally, Raghav Jain, Jhagrut Lalwani, Swapneel Mehta

Published: 2024-06-20 13:15:54+00:00

AI Summary

This research paper examines the threats of deepfakes to election integrity, evaluating the efficacy of existing detection methods. It provides an accessible summary for lawmakers and civil society actors to understand the technology and its implications for existing policies.

Abstract

It is becoming cheaper to launch disinformation operations at scale using AI-generated content, in particular 'deepfake' technology. We have observed instances of deepfakes in political campaigns, where generated content is employed to both bolster the credibility of certain narratives (reinforcing outcomes) and manipulate public perception to the detriment of targeted candidates or causes (adversarial outcomes). We discuss the threats from deepfakes in politics, highlight model specifications underlying different types of deepfake generation methods, and contribute an accessible evaluation of the efficacy of existing detection methods. We provide this as a summary for lawmakers and civil society actors to understand how the technology may be applied in light of existing policies regulating its use. We highlight the limitations of existing detection mechanisms and discuss the areas where policies and regulations are required to address the challenges of deepfakes.

Key findings

The paper highlights the limitations of existing deepfake detection mechanisms, particularly concerning low-quality videos and time consumption. It also emphasizes the need for improved authentication protocols, transparency requirements, media literacy programs, and legal frameworks to address the malicious use of deepfakes in elections.

Approach

The paper reviews existing deepfake detection techniques, categorizing them into spatial and spatio-temporal methods. It also analyzes the accuracy, usability, security, and computational efficiency of several deepfake generation tools, such as FaceSwap-GAN, CycleGAN, and StyleGAN.

Datasets

The abstract mentions the use of LinkedIn profile pictures in one example of deepfake detection (Sensity AI), but doesn't specify datasets used for the main evaluation of detection methods. Therefore, the datasets used are UNKNOWN.

Model(s)

The paper mentions several models including FaceSwap-GAN, CycleGAN, StyleGAN, and AttGAN, but doesn't detail the specific models used for the main deepfake detection evaluation. Therefore, the models used are UNKNOWN.

Author countries

UNKNOWN

← Previous