Harassment as Infrastructure: How Telegram's design enables TFGBV

AI Forensics analyzed 2.8 million messages across 16 Telegram groups and channels over a six-week period to study technology-facilitated gender-based violence in Italy and Spain, examining how the platform's design features enable organized abuse.

The investigation identified networks of nearly 25,000 users across the two countries engaged in spreading and monetizing non-consensual intimate images, including material meeting the definition of child sexual abuse material and depictions of incest and rape. The abuse was notably cross-border: 72% of Spanish-language content also circulated in Italian groups, and 8% of files appeared in both countries' groups simultaneously. Victims, predominantly women, were often identifiable through shared links to their real accounts. Perpetrators, typically young men, monetized the material through one-time fees of roughly €20-€50 or monthly subscriptions around €5, with nudifying bots used to fabricate synthetic abusive imagery at scale. TikTok, Instagram, and Reddit served as key pipelines funneling victims' content toward the Telegram groups.

The report concludes that Telegram's design enables this abuse infrastructure and calls for the platform to be designated a Very Large Online Platform under the DSA and for AI Act reforms addressing synthetic abuse content.