The cloud-based messaging platform Telegram is a hotbed for pornographic deepfakes, according to a safety investigation conducted by Sensity. The analysis describes the end-to-end encrypted messenger as a "deepfake ecosystem" where bots are able to turn everyday photos of people into "naked" content, especially of women.
These bots are powered by artificial intelligence, by Sensity, and perform “stripping” actions on a photo without clothes, often for the purposes of public humiliation and extortion. The company describes
- https://t.me/+NJ9dOO7okbMwZDQ0
- https://t.me/+D248nBps-zo2OTM8
- https://t.me/+9-HaKwj0_7I0Yzdk
- https://t.me/+ji5dVB_gocE5Zjhk
- https://t.me/+V6FoegpBESU2NDRk
- https://t.me/+UXVp9ifkA-ljMTY0
- https://t.me/+_O4b8Uc5cXowNTc8
- https://t.me/+V6FoegpBESU2NDRk
- https://t.me/+RRfnUf7uzJE1Yjhk
- https://t.me/+GI1Ne7GyVBo2ZmI0
- https://t.me/+c6qdQxeqVAczYWZk
- https://t.me/+E7vor0Hx8BE2NmM0
- https://t.me/+b2wiBTfae9Q3YTk8
this process as a kind of editing function that can be easily used by owners of smartphones and traditional computers. All they have to do is grab a photo from their personal image gallery and allow the bots to generate the deepfake. These photos can come from anywhere and everywhere on the Internet.
What Sensity found - Here are the main conclusions of Sensity:
- The company has discovered robots creating images that have targeted 104 women to date.
- These images were then propagated across the platform until the end of July of this year.
- In the past three months alone, those deepfakes have increased by 198%.
- At least 70 percent of those targeted have seen their poached images on social media or other private sources.
- Some of these victims were minors.
- Demographically, the company found that Telegram had received advertising from the Russian website VK while the deepfake bot garnered 101 people to participate. To exacerbate matters, bots operate for free. Users don't have to pay anything unless they want the watermark removed, which only takes $080 at most.
How do these bots work - In previous reports, Contribution explained how robots are able to generate their own images from original sources such as photos and videos. In the case of deepfake bots from Telegram, these bots rely on the previously mentioned generative adversarial networks. They strip down the original image of the clothed woman for a "realistic approximation" of the individual's anatomy.
This type of synthesis depends on the bot's ability to effectively guess the body parts and height of the individual. Of course, in order for this to work "perfectly", these robots are trained on large amounts of visual data related to women's bodies.
This is bad news for Telegram. The network already has a dubious reputation thanks to white supremacists frequently using the messenger. Now, with bots making fake nudes with unsuspecting people, including minors, the messaging program is only cementing its image as a deeply compromised platform.
