This is a blog post about our paper "From Underground to Mainstream Marketplaces: Measuring AI-Enabled NSFW Deepfakes on Fiverr" (USEC 2026, co-located with NDSS), by Mohamed Moustafa Dawoud, Alejandro Cuevas, and Ram Sundara Raman — a measurement study on AI-generated NSFW content on Fiverr, a mainstream freelance platform with 3.5 million active buyers.
What we found
We identified roughly 600 gigs advertising AI-generated NSFW content on Fiverr, despite the platform's explicit prohibition. The growth is striking: 74.9% of NSFW sellers joined Fiverr in 2025, compared to just 28% in mainstream AI categories.
Key findings include:
- 82.8% of listings offer deepfake-enabling features such as face-swapping or photorealistic imagery
- Roughly 20% offer custom AI model training, and 1.4% explicitly request photos of specific individuals
- Over half the gigs mention OnlyFans, with Instagram–OnlyFans combinations being the most common — suggesting Fiverr functions as upstream infrastructure enabling content that violates other platforms' policies
- NSFW sellers predominantly use open-source tools like Stable Diffusion and ComfyUI
- Western sellers (US and UK) comprise over 25% of this segment, despite both nations recently enacting deepfake legislation
What we recommend
We argue that platforms should implement predictable keyword-based detection, require consent verification for custom model training requests, and engage payment processors to apply pressure for compliance.
Read the full blog post on the RANDLab website, or see the full paper at NDSS.