Deepfakes are the result of human image synthesis created through artificial intelligence. The process involves combining and superimposing existing images and videos onto source images or videos using a machine learning technique known as generative adversarial network.
As the rise of deep-fakes emerges across the web, so do concerns over the potential use of them as tools in social engineering scams and cases of identity fraud.
Facebook is one social media platform looking to combat the problem through issuing a challenge titled the Deepfake Detection Challenge (DFDC). $10 million is being put towards the development of DFDC, and the launch and release of dataset is set to happen in December.
The format of the DFDC will be similar to that of a game in the sense that it will come with a leader board and prizes. Essentially, the challenge asks for users to search through a dataset of faces and videos – released by Facebook, to detect deepfakes. The dataset will consist of imagery of paid, consenting actors.
Facebook is particularly invested in taking stance against deepfakes for reasons that go back to situation that occurred in May, involving a deepfake of Mark Zuckerberg.
For the most part, I consider the DFDC to be an interesting concept as it is calling for social media users to play a role in detecting fakes over the internet. I want to believe that the incentive behind this challenge is to teach participants distinguish technological-manipulated imagery on the web from imagery that’s content is real, raw, and unedited. Hopefully, this challenge will at very least raise awareness and inform people of the existence deepfakes on the web.