Italian Authorities Investigate Deepfake Bots On Telegram

The Italian Data Protection Authority has started an investigation into the widespread use of bots that generate fake nude images on messaging app Telegram. The news follows an investigation by security firm Sensity, which found that as of July 2020 more than 100,000 faked images had been generated and shared in public Telegram channels.

The bots can generate fake nudes that have watermarks or that show only partial nudity, and users pay to “reveal” the whole image. Users could submit a photo of a woman to the bot and receive a version of the photo back with clothing “removed” and no indication that the image had been altered. And according to Sensity, a limited number of the bot-generated images, most of which are pulled from social media accounts and then manipulated, are of victims who “appeared to be underage.”

Sensity found that the fake nudes were generated by the DeepNude software, which surfaced online last year. Its creator took down the website saying “the probability that people will misuse it is too high.” According to Sensity, however, the software has been reverse-engineered and is widely available on open source repositories and torrenting websites.

“The ease of use of this program makes anyone with a photo on the web potentially victims of deep fakes,” the Italian agency said in a statement Friday (in Italian) announcing its investigation into the matter. It plans to ask Telegram to provide information to verify whether it’s complying with data protection regulations, according to its statement.