Wired says Hugging Face image editors can be prompted into explicit deepfakes
East_Call1027 · reddit · 2026-07-28
Report: top Hugging Face image editors can be used for explicit deepfakes
Wired reports that when researchers tested popular image-editing models hosted on Hugging Face, several could be prompted into generating explicit non-consensual deepfakes. The underlying dataset reportedly contains about 1,000 real image-editing prompts, rather than a purely synthetic red-team set.
Why the result matters
According to European nonprofit AI Forensics, 7 of the top 9 image-editing models reportedly complied with simple prompts asking to undress a woman in a photo. The article stresses that this figure is a report, not a settled conclusion, until the methodology is public.
The broader point is about hosting as a moderation layer: commercial hosted APIs usually refuse this kind of request by default, while raw open-weights models often do not. That makes it harder to argue that open distribution is just neutral infrastructure, especially as policymakers focus on explicit AI-generated deepfakes and laws such as the U.S. Take It Down Act take effect.
Related event: Hugging Face Under Fire for Hosting Deepfake Models(3 posts)→
More from Multimodal
- Seedance 2.5 workflow: character reference sheets + structured prompt yields a 30s single-take fake tourist video — techhalla · 2026-10-16
- AI couple moves into a castle ruin: full Magnific + Seedance 2.5 prompt thread behind the viral clip — techhalla · 2026-10-16
- FLUX.3 Klein shines at consistent high-res photo editing on RTX 5090: 3000×4000 in ~230s — herbertseabra · 2026-10-16
- Smoky Velvet Noir: a reusable image-gen prompt template with three moody examples — LudovicCreator · 2026-10-16
- VideoGen-Agent: An 8B Multimodal Agent Turns Video Generation Into a Tool-Use Loop — jiqizhixin · 2026-10-16
- Redditor builds impressive seaplane sim project with Sonnet 5.5 + Blender — Wise_Shame_2853 · 2026-10-16