Artificially generated ‘deepfake’ video, image, audio and text pose a major challenge for governments, politicians, companies, CEOs, celebrities, academics and commentators.

A growing number of technology companies, academic researchers and NGOs are developing ways to identify false and misleading AI-generated media. Progress is being made, but there is a long way to go until a viable, comprehensive solution, or set of solutions, are available.

Here is a list of tools, datasets and toolkits on the market and in the pipeline:

Adobe Photoshop Face Aware Liquify detectorAdobe/UC BerkeleyPhotoshop imagedetection
ADVBOX (pdf)Baidu X Labmultipledetection
Amber Detect Ambervideodetection
AssemblerJigsaw/Google ResearchStyleGAN video detection
DeeperForensics 1.0 (pdf)SenseTime Research/Nanyang Technological Universityfacial imagedataset
Deepstar ZeroFoxvideotoolkit
DeeptraceDeeptrace Labsvideomonitoring, detection, mitigation
DeepFake DetectionDessavideodataset
Deep VoicePindropvoicemonitoring, detection, mitigation
DiggerDW Innovation/iLab ATC/Fraunhofer IDMTaudiodetection
FaceForensics++ Technical University of Munich/ University Federico II of Naplesfacial imageanalysis
Grover Allen Institute for AInewsdetection
GLTRMIT/IBM Watson AI Lab/Harvard NLPtextdetection
Media ForensicsDARPA MediForimage, videodetection
Reality DefenderAI Foundationimage detection
ResemblyzerResemble AIaudiodetection
SherlockSherlock AIvideodetection
Trusted Media CaptureSerelayimage, videodetection
Truepic VisionTruepicphoto and videodetection
Visual deepfakes Jigsaw/Google Research imagedataset

The list above is not comprehensive. I’d be interested to hear of any tools or toolkits you recommend.

Check out my selection of white papers, frameworks, commentary and other resources exploring AI and machine learning from a risk management, reputation management and communications perspective.

Last updated February 24, 2020

%d bloggers like this: