DeepNude was a deepfakes application which allowed to alter photos of a person to make them appear nude, designed to work on females. Several days after its June 23rd, 2019, release and rapid climb to popularity, the app was permanently shut down over the concerns of potential misuse, although already downloaded copies of the software were continued to be shared.
On June 23rd, 2019, Windows and Linux application DeepNude, developed by an anonymous person known only as Alberto, was made available for download online (website no longer available).[1][2]
The software, based on a neural network, automatically edited photographs of people to make them appear fully nude. The app only had support for female naked bodies, although an update or a separate app for male bodies was also planned.[3]
On June 27th, 2019, Vice Motherboard featured a story on the app, titled "This Horrifying App Undresses a Photo of Any Woman With a Single Click", also talking to the creator of the app.[3] In the following days, multiple news outlets covered the story, including articles by The Verge,[4] Vox[5] and CNET.[6]
On June 27th, the official website experienced a series of crashes due to high number of visitors.[7] On the same day, the creator of the DeepNude announced that the service had been permanently shut down due to concerns of potential misuse of the application.[8]
DeepNude Shutdown Announcement
On the same day, an unknown Reddit user posted a link for a cracked version of the app.[9]
In early July 2019, GitHub removed submissions of DeepNudes code from the repository. On July 9th, the technology news site Motherboard published an article about the removal, which included a statement from a GitHub spokesperson:
"We do not proactively monitor user-generated content, but we do actively investigate abuse reports. In this case, we disabled the project because we found it to be in violation of our acceptable use policy. We do not condone using GitHub for posting sexually obscene content and prohibit such conduct in our Terms of Service and Community Guidelines."
DeepNude uses a pix2pix open-source algorithm to analyze the input image, identify clothing and remove it, replacing it with a generated image of a naked body. The algorithm was trained on a dataset of over 10,000 nude photos of women.[3]
The free version of the app had the output images partially obscured with a large watermark, removed in the $50 paid version. Paying additional $50 removed a smaller "FAKE" watermark (examples shown below, left and right).[10]
