Why I Go Looking For Creeps Online

I have been trying to answer this question for a while now.

In partnership with

A few weeks back, I got an email from a professor at NYU. She had read my AI undressing tools investigation, the one where we mapped the global supply chain behind these tools, and she had a question about a choice I had made. How did I decide to follow the perpetrators and the money, when most of the conversation around deepfakes stays with the victims?

It’s a fair question, and one I’ve been trying to answer for myself long before I started working on the story.

I’ve been reporting on gender for about fifteen years now, which is long enough to know that the story hasn’t changed as much as we’d all like to pretend. Consent was not understood then; it is not understood now. A photograph could ruin you, when the images were real. Women left careers, left platforms, left public life. AI just made all of it worse: faster, cheaper, more accessible. A screengrab and sixty seconds with a free tool can now generate what used to take a vindictive ex and a Photoshop licence.

The impact on women hasn’t changed. What keeps changing is everything else: the person doing it, the tools they use, how they get paid. 

Kate Del Castillo Netflix GIF

Gif by ammejia on Giphy

The perpetrator shifted from the creepy boy in class to a mobile phone shopkeeper to an engineering student at Vellore Institute of Technology, one of India’s top engineering colleges, building an app to auction Muslim women for speaking up on Twitter. The tools went from threatening emails to Photoshop to apps freely available on the Play Store. The payments went from cash to crypto to UPI to platform credits bought through Indian middlemen. All of this keeps evolving, and nobody is really mapping the evolution.

I call this the accountability pipeline. Not just what happens to women, but who did it, with what, and whose money paid for it. Those are the parts that keep changing, and those are the parts where you can actually intervene. Danielle Citron called it whack-a-mole years ago, and she was right. If you only go after the apps, another one appears. You have to follow the infrastructure: the payment processors, the app stores, the platforms hosting the models.

When I pitched the story to the Pulitzer Center, the idea was to find the Indian networks behind AI undressing bots. What we found was bigger. A supply chain that starts in China, where the base models are built, routes through Europe, runs on American hosting platforms, and ends with Indian users, Indian payment agents, and Indian women caught in the middle. At every point, money changed hands. At every point, someone could have intervened. 

So I spent months inside Telegram groups and online marketplaces where this content is sold and distributed (about as pleasant as it sounds), tracking who supplies it, who buys it, how the money moves. That is how I found Pratim Bora’s networks.

Bora had created an AI-generated character, Babydoll he called her, modelled on a real woman from Assam. For five years, he sat in his parents’ home, making Babydoll more convincing, more explicit, more profitable. In 2023, someone contacted him promising extra bonuses for extra content, two years before anyone knocked on his door. His plan had long shifted from revenge to revenue. He was jobless, living at home, making sure his parents didn’t find out. His father, who sells homemade pickles and milk from a small shop, had no idea what his son was doing from inside the house for years. He told us he thought his son was doing a work-from-home gig.

The media had reported on the AI-generated character as though she were a real woman. That reportage made the neighbours believe it was real, so she filed a complaint. Police traced Bora through Meta and arrested him. That was the intervention, at the very end, after five years of content being created and distributed, of payments flowing through crypto and UPI, of platforms hosting the tools he used. Five years, and the system caught one man at the bottom of the chain.

A year and a half on, the trial hasn’t started and the content is still up. India’s IT Rules now require a two-hour takedown for non-consensual intimate imagery, which is something. But there is still no standalone deepfake law, no government data on non-consensual synthetic imagery.

When he was arrested, he told the police he had made a terrible mistake. He wanted to apologise. That realisation, perhaps, may have come sooner if someone, somewhere in that five-year pipeline, had been held accountable before him.

On Monday, the Centre for Communication Governance at NLU Delhi launched Illustrated Fakely, their report on non-consensual synthetic imagery. The panel I was on focused on consequences for victims, and those consequences are exactly where the other accountability questions live. 

In the Sulli Deals case, Muslim women were listed for auction precisely because they were Muslim and outspoken. The woman in the Assam case had political aspirations; now she doesn’t step out of her house. Religion, caste, where you live, your family, all of it compounds what happens next. For a story on Telegram CSAM networks a few years ago, I spoke to a young woman whose non-consensual video was shared when she was seventeen. She is nineteen now, multiple cases filed, her videos still circulating. Her parents took her phone and pulled her off social media. Each of these stories tells you where the system failed: no data collected, no reporting framework, no support structure. 

During the discussion, Anita Gurumurthy of IT for Change asked me: what can change?

And I went back to Bora. To the five years nobody was watching. To the supply chain that made it possible. To the money.

I’ve been doing this long enough to know I’m not going to single-handedly reform society with my journalism. No matter how many reporting trips, how many panels, how many newsletters.

But I am hopeful about the immediate things. Take the content down. Take the payments down. Find the networks. Follow the money backward. And maybe, the next time there is a government discussion on India’s AI moment, someone will have the data to show what that moment costs women.

So, to answer the professor: I followed the perpetrators and the money because after fifteen years, I know what happens to the women. What I need to keep understanding is who is profiting. That is the other story, and the one where something can change.

On My Bookmarks

Cloaked Network

WIRED mapped over 400 apps disguised as pet portrait generators and workout coaches that secretly hide nudify tools behind a server-side switch, all linked to Chinese language developers and advertised through thousands of explicit ads on Meta’s platforms.

Tiny Thing

Gideon Lewis-Kraus wrote in The New Yorker on the semiconductor, the thing most people never think about. The thing that powers your phone, your car, the AI boom, and an intensifying geopolitical contest over who gets to make them.

What Is This Oops GIF by Blake Lively Fan

Gif by blakelivelyfan on Giphy

The Opt-Out Question

 Bridget Todd, Douglas Rushkoff and others on the Tech Policy Press podcast wrestle with whether it still makes sense to stay on the internet, or whether the cost of platform dependence has gotten too high for what we actually get back.

Got a story to share or something interesting from your social media feed? Drop me a line, and I might highlight it in my next newsletter.

See you in your inbox, every other Wednesday at 12 pm!

MESSAGE FROM OUR SPONSOR

It's time to set up your holiday streaming campaign

Are you ready to reach holiday shoppers on CTV? For many streamers, the season is already underway.

Last year, according to Roku internal data, holiday content searches on Roku saw an 86% increase during October. And in November, those searches rise an average of 39% weekly. This surging holiday spirit is mirrored by consumer activity, as about half of Roku streamers shopped Black Friday and Cyber Monday deals in 2025.

Meeting those shoppers is easier than you might think. Begin with the seasonal outcome you hope to drive: awareness, online sales, leads, store visits, or another measurable goal. From there, define your audience, upload your creative, set the budget and campaign dates, and establish measurement before launching. You can even create a shoppable ad unit that lets streamers buy with a click of their remote.

Our holiday playbook offers a month-by-month guide to reaching shoppers during this all-important time of year.

First-time spenders get a $5,000 match on the first $5,000 spent with code BEEHIIV5K. Terms apply.

Was this email forwarded to you? Subscribe