Discussion Paper: The Threat of Real Time Deepfakes

View on arXiv ← Back to list

Authors: Guy Frankovits, Yisroel Mirsky

Published: 2023-06-04 21:40:11+00:00

AI Summary

This discussion paper highlights the emerging threat of real-time deepfakes, which enable attackers to create realistic audio and video impersonations in real time for social engineering attacks. The authors discuss the challenges in defending against these attacks and suggest a shift towards active and out-of-band defense strategies.

Abstract

Generative deep learning models are able to create realistic audio and video. This technology has been used to impersonate the faces and voices of individuals. These ``deepfakes'' are being used to spread misinformation, enable scams, perform fraud, and blackmail the innocent. The technology continues to advance and today attackers have the ability to generate deepfakes in real-time. This new capability poses a significant threat to society as attackers begin to exploit the technology in advances social engineering attacks. In this paper, we discuss the implications of this emerging threat, identify the challenges with preventing these attacks and suggest a better direction for researching stronger defences.

Key findings

Current deepfake detection methods are largely ineffective against real-time deepfakes due to limitations in practicality, media quality, and delivery methods. The authors advocate for a move towards active and out-of-band defense strategies to counter this evolving threat. They highlight the need for further research into these alternative approaches.

Approach

The paper focuses on discussing the limitations of current deepfake detection methods, which mainly rely on analyzing content for artifacts. It proposes a shift toward active defenses (e.g., CAPTCHAs) and out-of-band defenses (e.g., source verification) that do not depend on analyzing the media content itself.

Datasets

UNKNOWN

Model(s)

UNKNOWN

Author countries

Israel

← Previous