AI Generated Fake Audio as a New Threat to Information Security: Legal and Forensic Aspects

Elena Galyashina, Vladimir Nikishin

2021

Abstract

A speech pattern synthesis is the classic task for AI which goal is to automate the process of text sounding with natural voice characteristic features. The corresponding sets of techniques are called voice cloning that could be relatively easy recognized by listeners as a machine speech. But in the past two years the new developments in the area of AI and deep generative networks have significantly improved the quality of speech synthesis and brought to life the technology of deep fake audio that appear rather authentic. Cyberattacks using convincing sounding models successfully emulating people’s voices have escalated in number, frequency, and impact, drawing increased attention to the vulnerabilities of cyber systems and the need to increase their security. False audio content generated with neural networks has become difficult to be detected that causes a widespread distrust of audio (voice) evidence and brings a serious societal harm. In the face of this threat, there is a significant concern and interest among forensic speech researchers and the juristic public about the malicious implications of AI and risks of fake audio attacks for cybersecurity. This article describes some legal and forensic aspect of exposing fake voice audios generated with neural networks.

Paper Citation

in Harvard Style

Galyashina E. and Nikishin V. (2021). AI Generated Fake Audio as a New Threat to Information Security: Legal and Forensic Aspects. In Proceedings of the International Scientific and Practical Conference on Computer and Information Security - Volume 1: INFSEC, ISBN 978-989-758-531-9, pages 17-21. DOI: 10.5220/0010616700003170

in Bibtex Style

@conference{infsec21,
author={Elena Galyashina and Vladimir Nikishin},
title={AI Generated Fake Audio as a New Threat to Information Security: Legal and Forensic Aspects},
booktitle={Proceedings of the International Scientific and Practical Conference on Computer and Information Security - Volume 1: INFSEC,},
year={2021},
pages={17-21},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010616700003170},
isbn={978-989-758-531-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the International Scientific and Practical Conference on Computer and Information Security - Volume 1: INFSEC,
TI - AI Generated Fake Audio as a New Threat to Information Security: Legal and Forensic Aspects
SN - 978-989-758-531-9
AU - Galyashina E.
AU - Nikishin V.
PY - 2021
SP - 17
EP - 21
DO - 10.5220/0010616700003170