The president of the United States is arguably one of the most powerful people in the world. Given the weight his word carries, one would hope that American citizens could reliably recognize him when he speaks. But as the capacity to fake videos gets more sophisticated, Matt Groh wanted to test how well people were able to identify political deepfakes.
In a peer-reviewed paper in Nature Communications, Groh and his coauthors Aruna Sankaranarayanan, Nikhil Singh, Dong Young Kim, Andrew Lippman, and Rosalind Picard from the MIT Media Lab describe a dataset they created for studying deepfakes, which consists of 32 short speeches by President Joe Biden and former President Donald Trump.