Detecting DeepFakes – “Seeing Strings”
U.S. flag

An official website of the United States government

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

Focus Area:

Influence and Information Capabilities

Project Cost:

$3,000,000

Project Duration:

18 months

Performer:

Los Alamos National Laboratory

Deliverable:

A benchmarked, standalone, updateable prototype app, which can be used by the USG and select partners in near real-time, while limiting the ability of an adversary to compromise our detection capabilities in the future.

End User:

EUCOM J2 and J3

Description:

“DeepFakes” are media files that have been created or edited using a new type of Artificial Intelligence (AI)-assisted video, image, and audio generation algorithm. These hyper-realistic, constantly evolving, digital forgeries use sophisticated Machine Learning (ML) techniques to produce convincing but false video depictions of individuals doing or saying things they never did, without their consent or knowledge. DeepFakes have the potential to disrupt every facet of our society, triggering dangerous international, domestic, and personal consequences. In addition, the use of AI algorithms to automate much of the creation and editing process means that this technology is widely available.

Operational Impact:

The application provides support for analysts in the detection and identification of puppet-and-driver DeepFakes used in adversarial mis/disinformation campaigns.

Transition:

The app has transitioned to EUCOM.