The Silicon Valley Leadership Group, in partnership with the CITRIS Policy Labs at the University of California, Berkeley, announced that Katy Jiang, ’21 MS Software Engineering, won the 2020 Deepfake Education Competition with her three-minute explanatory video. The competition challenged students of all levels to create engaging video content to educate the public on deepfakes, the use of artificial intelligence to manipulate images and video to influence public opinion.
“Deepfake is a form of artificial intelligence. The word deepfake combines deep learning and fake,” said Jiang. “It can produce a persuasive counterfeit by studying photographs and videos of a target person from multiple angles, and then mimicking its behavior and speech patterns. By making this video, I want to educate people about the technology behind their day-to-day entertainment application and raise people’s awareness that deepfakes can also be used in committing crimes such as frauds and scams.”
A 2019 Pew Research study reported that two-thirds of Americans (67 percent) say made-up news and information cause a great deal of confusion. Associate Chair of the Computer Engineering department Magdalini Eirinaki, who is teaching Jiang’s Web and Big Data Mining course this semester, recommended that Jiang submit her video, originally a class assignment, for the Deepfake Education competition.
“Identification and spread of fake news (whether in text or deepfakes) has been on my radar as a very interesting and critical research problem,” said Eirinaki. “This technology can be very easily weaponized and used to enhance the perceived credibility of fake news and disinformation campaigns. This can have even more devastating effects than current fake news, ranging from politics, to the environment (e.g. global warming), to public health (e.g. spreading disinformation about the COVID-19 pandemic). It is therefore more critical than ever for the research community to develop more sophisticated techniques to keep up with the deepfake technology to promptly identify and remove/flag them before too much harm is done.”
Jiang’s video explains how deepfakes can be used to impact how people interpret fake news—especially timely during a hotly contested presidential election and a global health pandemic. Using engaging visuals, music and voiceover to describe the dangers that deepfakes pose to democracy, she encourages viewers to assess content carefully before sharing it on their social media platforms. She demonstrates how many free apps are available to superimpose public images with false or misleading suggestions, inserting her own face over that of celebrities and politicians. Her winning submission will be featured on CITRIS media channels and through the Silicon Valley Leadership Group’s media channels. The recognition also includes a $2,500 prize.
“As we are in the pandemic and the election is coming, deepfakes pose a danger to democracy,” said Jiang. “Fake news will influence everything from stock prices to the election. People should be critical about what we see online.”