'Deepfake' of 'drunk' Nancy Pelosi is actually a cheapfake
An edited video of House Speaker Nancy Pelosi, in which she appears to be drunkenly slurring her words, making her look incompetent, went viral on social media.
The video was mistakenly labelled as a deepfake, but is a cheapfake: a crude manipulation which does not use AI but instead simple video editing tools. In this case, the video was slowed down around 75% of the original speed to make Pelosi’s voice sound garbled and warped and then altered audio made her voice sound a ‘normal’ pitch.
Both president trump and Rudy Giuliani shared the video on Twitter as 'fact'. The altered video has been viewed millions of times, attracting comments speculating on Pelosi’s health, supposed use of drugs, and other apparent ailments. Highlighting the struggle to deal with disinformation, social media platforms continued hosting the altered video due to lack of regulations. Despite the apparently malicious intent of the video’s creator, Facebook said it will only downgrade its visibility in users’ newsfeeds and attach a link to a third-party factchecking site pointing out that the clip is misleading.
We are yet to see sophisticated deepfakes in US or UK politics. Potentially because cheapfakes, made with basic, cheap (as the name implies) and readily available video editing software, create a big enough impact on their own. Fake videos are most effective if they’re timed for maximum chaos, close to Election Day when newsrooms won’t have the time to investigate or debunk them. There are major calls for safer regulation to be put in place, especially with cases of political misuse, disinformation, fraud and harassment.
Several people reported this case (and still continue to) as a deepfake causing an echo chamber online both on social media and the news as information cascades continued to spread misinformation.