R Street Institute
R Sheets Technology and Innovation

R Sheet on Deep Fakes

by Jeffrey Westling AND Charles Duan
March 5, 2019
Download PDF Print

Authors

Jeffrey Westling
Former Resident Fellow, Technology & Innovation
Charles Duan
Charles Duan
Former Senior Fellow

Key Points

Deep fakes are a class of AI-generated audiovisual material designed to appear realistic.

Because of how closely they mimic reality and the ease with which a bad actor can utilize the technology, some have begun to worry about the harms that the technology can do to society.

However, deep fakes are merely the most recent in a long line of deceptive technologies.

Regulators should not use the fear of deep fakes as an independent justification for regulating the more general problem of disinformation.

Background

The term “deep fake” refers to AI-generated audiovisual media that can convincingly resemble authentic content. Although realistic animated video has been a Hollywood mainstay for decades, the concept has attracted attention recently due to advances in AI technology that significantly simplify the ability to generate realistic representations. Specifically, “deep learning” algorithms (hence the name “deep fakes”) are designed to teach computers how to create fake videos that are virtually identical to real ones (at least to the naked eye).

Read the full study here.

Featured Publications