WHAT YOU NEED TO KNOW:
- Researchers from Sensity security firm discovered a “deepfake ecosystem” on the Telegram app where bots are programmed to generate fake nude images of women.
- A software called DeepNude is being used to generate the nude images.
- Bots are also programmed to handle payments that generate revenues collected by their creators.
A “deepfake ecosystem” on the Telegram messaging app was discovered by researchers, based on a report. Generated by bots, massive fake nude images of women have spread on the said app.
According to users who were interacting with these bots, they focused on making fake nudes of women they knew through posted images from social media sites.
Security firm Sensity, the investigating team, has described the spread of deepfakes as “visual threat intelligence.” The firm’s researchers discovered over 100,000 images that were generated and shared in public Telegram channels until July 2020.
Sensity found that a large majority of users from these channels came from Russia and nearby nations. Many of these channels were still active as verified by The Verge.
With free of charge, the bots could develop fake nudes but with watermarks or limited nudity. To make a fully nude image, users would need to pay at least a few bucks. A beginner rate, for instance, would cost around 100 rubles fee ($1.28) that could generate 100 fake nudes without watermarks for a span of one week.
The bots are also being utilized to handle the fees as their creators make a living out of it.
The software, known as DeepNude, is being used to generate the images. The first nude photo was shown on the web last June but was quickly deleted after gaining mainstream media attention, saying “the probability that people will misuse it is too high.”
DeepNude, though, continued to spread across backchannels as Sensity said that it “has since been reverse engineered and can be found in enhanced forms on open source repositories and torrenting websites.”
To generate fake nudes, the software uses an artificial intelligence method called generative adversarial networks or GANs. Most images are easily seen as altered while some could be treated as real photos.
According to Sensity, it is “reasonable to assume” that most users “are primarily interested in consuming deepfake pornography.”
Giorgio Patrini, Sensity’s CEO and co-author of the report, told The Verge that the “key difference is the accessibility of this technology.”
He continued: “It’s important to notice that other versions of the AI core of this bot, the image processing, and synthesis, are freely available on code repositories online. But you need to be a programmer and have some understanding of computer vision to get them to work, other than powerful hardware. Right now, all of this is irrelevant as it is taken care of by the bot embedded into a messaging app.”
Both Sensity and The Verge have not yet received a response from Telegram when they requested for a comment on the issue.
Source: The Verge