On the Horizon: Interactive and Compositional Deepfakes

View on arXiv ← Back to list

Authors: Eric Horvitz

Published: 2022-09-05 01:08:05+00:00

AI Summary

This research paper explores the emerging threats of interactive and compositional deepfakes, which pose significant challenges to discerning fact from fiction. Interactive deepfakes enable realistic impersonation through multimodal interaction, while compositional deepfakes integrate multiple deepfakes and fabricated events to create persuasive synthetic histories.

Abstract

Over a five-year period, computing methods for generating high-fidelity, fictional depictions of people and events moved from exotic demonstrations by computer science research teams into ongoing use as a tool of disinformation. The methods, referred to with the portmanteau of deepfakes, have been used to create compelling audiovisual content. Here, I share challenges ahead with malevolent uses of two classes of deepfakes that we can expect to come into practice with costly implications for society: interactive and compositional deepfakes. Interactive deepfakes have the capability to impersonate people with realistic interactive behaviors, taking advantage of advances in multimodal interaction. Compositional deepfakes leverage synthetic content in larger disinformation plans that integrate sets of deepfakes over time with observed, expected, and engineered world events to create persuasive synthetic histories. Synthetic histories can be constructed manually but may one day be guided by adversarial generative explanation (AGE) techniques. In the absence of mitigations, interactive and compositional deepfakes threaten to move us closer to a post-epistemic world, where fact cannot be distinguished from fiction. I shall describe interactive and compositional deepfakes and reflect about cautions and potential mitigations to defend against them.

Key findings

The paper highlights the increasing sophistication and potential for malicious use of interactive and compositional deepfakes. It emphasizes the need for technological, policy, and educational solutions to counter the threats posed by these advanced forms of disinformation. These solutions include improved media literacy, digital content provenance methods, and robust detection techniques.

Approach

The paper analyzes the technical advancements in generative AI, multimodal interaction, and causal modeling that enable the creation of interactive and compositional deepfakes. It explores the potential for automation and mixed-initiative control in these deepfakes and discusses the challenges of detecting and mitigating their use in disinformation campaigns.

Datasets

UNKNOWN

Model(s)

Generative adversarial networks (GANs), WaveNet, multimodal neural models, and other deep neural models for speech recognition, speech production, and facial expression generation are mentioned as underlying technologies, but no specific models used for detection are described.

Author countries

United States

← Previous