The Internet is home to at least 14.678 deepfakes, according to a new report from DeepTrace, a company that builds tools to identify synthetic materials. But most of them were created not to spoil the elections.
Top: Deepfakes arrived on the scene at the end of 2017. The word was used to describe the false pornography generated by AI, with the heads of actresses on the bodies of the adult film stars. Since then, the sense of “deepfakes” has expanded to refer to any type of manipulated video, as Mark Zuckerberg with a fake speech about Facebook. This has stoked fears about the end of the truth and potential deepfakes to swing elections.
The Internet is for porn: The report found that most videos are not trying to influence politics. At least 96% of deepfakes were still good old fake porn. “Pornography Deepfake is a phenomenon that exclusively targets women and night,” the authors write. All contained fake porn women, most famous actresses and musicians. (Other deepfakes nonpornographic YouTube included most men.) The number itself is not high, but it is worrying that it is growing so quickly.
- South Park Has Been Banned In China For Their Joke
- No! More Rape – Exclamation points kill! Exclamation points! Kill
Holding back the law: The issue has attracted the attention of legislators. In California, Governor Gavin Newsom has signed into law two bills that limit what people can do with deepfakes. A law making it illegal to distribute malicious deepfake a politician within two months of an election. (The ACLU and the Electronic Frontier Foundation have pushed back, saying the law is too broad and ill political speech.) The second deepfake law approximates how the manipulations are actually used. It allows people to sue if their image is used in deepfake sex without consent.
According deepfakeapp on imgur, anyone can download and run FakeApp can create one of these videos with just one or two high quality face fake videos they want. The wiki subreddit FakeApp says is “a community developed desktop application to run the algorithm deepfakes without installing Python tensorflow, etc.,” and that all you need to run it is a “good GPU [unit graphics processing, the kind that sets high-end 3D video require] with CUDA support [parallel computing platform from NVIDIA and the programming model]. “If users are not the appropriate GPU, they can also rent cloud graphics processors through services such as Google Cloud Platform. Running the entire process from data extraction to Image conversion picture of one side over another, would take about eight to 12 hours if it is done correctly. Other people reported spending much longer, sometimes with disastrous results.