McAfee, the device-to-cloud cybersecurity company, today announced the launch of the McAfee Deepfakes Lab to provide traditional and social media organizations advanced Artificial Intelligence (AI) analysis of suspected deepfake videos. These videos could be used to spread reputation-damaging lies about individuals and organizations during the 2020 U.S. election season and beyond.
“This election season brings with it a variety of critical concerns around election security and disinformation,” said Steve Grobman, CTO and SVP at McAfee. “Advanced AI technologies are making it possible for a surprisingly broad field of potential bad actors to create deepfake videos that portray candidates as saying and doing things they never did. Such bogus video footage could be released in the days leading up to November 3rd and influence voters in a particular direction. The McAfee Deepfakes Lab applies McAfee’s own AI capabilities to help media outlets identify and counter disinformation before it can inflict damage and sway voters.”
Deepfake videos are created by the superimposition of existing images, audio, and videos onto source media files by leveraging an advanced deep learning technique called generative adversarial networks (GANs). These GANs can be used to synthesize artificial videos that are indistinguishable from authentic ones. Unfortunately, they could be used to develop damaging, politically motivated video content that could conceivably spread across social media before implicated individuals and reputable news organizations could challenge the video’s claims.
The McAfee Deepfakes Lab will leverage the company’s data science expertise and tools that combine computer vision and deep learning techniques to exploit hidden patterns and detect manipulated video elements that play a key role in authenticating original media files. News organizations and social media platform managers will be able to submit suspected videos to the Deepfakes Lab and receive detection scores and heatmaps that explain McAfee’s assessment of the videos’ authenticity.
To engage the services of the McAfee Deepfakes Lab, media organizations may submit suspect video for analysis by sending content links to media@mcafee.com.
“Unfortunately, we may find ourselves in a prolonged ‘deepfakes race’ where AI advances make it easier to create these videos and compel us to apply as much self-education as technology to contain this form of disinformation,” Grobman continued. “We all need to become as savvy to deepfakes as we now are with ‘photoshopped’ images of celebrities doing inconceivable, out-of-character things. When we see something odd, question its source, recognize it, and, in doing so, inoculate ourselves to blunt the damage it can do to individuals and society. In the meantime, the McAfee Deepfakes Lab will help us keep pace with any bad actors seeking to manipulate us.”
For more information:
- Technology – McAfee Deepfakes Lab: Detecting and Defending Against Deepfakes
- Voter Tips – How to spot phony Deepfake videos this selection
- Video – McAfee CTO Steve Grobman discusses the Deepfakes threat to Election 2020
