AI’s dark marketplace: How Telegram bots monetize abuse at scale

A study of 2.8 million Telegram messages in Italy and Spain didn’t just document abuse—it mapped an entire supply chain. Nudifying bots strip clothing from images in seconds, deepfake tools fabricate consent, and automated archives organize victims’ likenesses like inventory. The ecosystem doesn’t just enable non-consensual intimate imagery; it optimizes it, with monetization baked into the workflow.

The mechanics are brutally efficient. A user uploads a photo, a bot processes it, and within minutes, the result is filed into searchable databases—or sold outright. According to available information, these tools lower the barrier to entry so dramatically that abuse becomes a volume business. No dark web required: it’s all happening in plain sight on Telegram, where channel operators profit from subscriptions, paywalled archives, or even blackmail schemes.

What’s striking isn’t just the scale, but the normalization. The study, published by The Decoder, frames this as an infrastructure problem—not a series of isolated incidents, but a system designed to extract value from violation. The real signal here is how seamlessly AI folds into the process, turning what was once labor-intensive exploitation into something as frictionless as ordering takeout.

The winners in this economy are the usual suspects: bot developers who pocket subscription fees, channel admins who monetize access, and the platforms that—despite repeated warnings—still host the activity. The losers are anyone whose image gets fed into the machine, but also the broader public grappling with the erosion of consent as a concept. When AI can generate convincing fakes from a single photo, the idea of ‘opt-in’ becomes a relic.

Early signals suggest this isn’t confined to Italy and Spain. The tools are language-agnostic, the payment methods borderless, and the demand global. Some users report seeing similar bots in Latin American and Southeast Asian networks, hinting at a template that’s easily replicated.

The ethical tension isn’t just about the technology—it’s about the platforms that profit from engagement while outsourcing moderation to underresourced teams or, in Telegram’s case, a hands-off philosophy that treats all content as equal.

For all the noise about AI’s creative potential, the actual story is how efficiently it serves the basest markets. The same tools that power art generators and chatbots also power abuse-as-a-service, with the added bonus of plausible deniability. (‘It’s just code!’) That’s just another way of saying the internet’s next gold rush won’t be built on ads or data—it’ll be built on exploitation, automated and scalable.