Deepfake detector wins PennApps XX

An app designed to detect deepfakes took home the grand prize at PennApps XX, beating nearly 250 tech projects developed over the course of a weekend. The top three team members took home a bevy of tech items, software subscriptions, and gift cards; nearly $100,000 worth of prizes were award across a variety of technical categories.

Team DeFake: Aarati Srikumar, Daniel Li, George Diwan and Sofiya Lysenko
Team DeFake: Aarati Srikumar, Daniel Li, George Diwan and Sofiya Lysenko (Image: Penn Engineering)

PennApps, the nation’s first student-run college hackathon, was founded in 2009 and has been running biannual invitational events since. Contestants from around the world arrive on a Friday afternoon, form teams and come up with ideas for digital apps or hardware projects, which they must complete before a demo session on Sunday morning. There, the projects are evaluated by a panel of tech industry judges, who rate them on originality, technical difficulty, polish, and usefulness.

This year’s winning project was DeFake, a browser extension that uses machine learning to assess the likelihood that a given video has been subtly manipulated. Deepfakes are a new method of using artificial intelligence to impersonate famous figures, using pre-existing video and audio clips to generate photo-realistic fabrications. Worried about deepfakes being “particularly influential in the outcome of this and future elections,” the team “wanted to develop a level of protection and reality for users.”

Read more at Penn Engineering.