gen‑ai.news
← Back
Multimodal

K-pop Fans Are Calling Out Creepy Deepfakes of Idols

K-pop Fans Are Calling Out Creepy Deepfakes of Idols

Within K-pop fan communities, a growing number of people are speaking out against the creation of AI-generated images and videos that depict idols in sexualized scenarios. The content typically uses deepfake techniques - training or prompting generative models on an idol's likeness to produce material the person never consented to. As these tools have become more accessible, the volume and realism of such content has increased noticeably.

What makes this situation somewhat distinct from other deepfake controversies is that the opposition is coming largely from inside the fanbase itself. Fans are flagging content on social platforms, compiling documentation, and in some cases directly confronting the accounts responsible. This kind of internal community policing is notable because fan spaces have historically struggled to self-regulate, particularly when the behavior in question is framed by some participants as a form of creative expression or fandom engagement.

The idols targeted are real people with public profiles, and the harm from non-consensual sexualized imagery is well established regardless of whether it is photographic or AI-generated. Several countries have moved to update laws around synthetic media to cover exactly this kind of material, though enforcement remains inconsistent, especially across borders. South Korea has been among the more active jurisdictions in addressing deepfake abuse, having strengthened relevant provisions in recent years, but the content often originates and circulates across multiple platforms and regions.

The broader pattern here is not unique to K-pop - athletes, actors, and other public figures face similar problems - but the intensity of K-pop fandom culture, combined with the youth of many idols and their highly visual public presence, makes the issue particularly acute in this space. The fan-led response, while unlikely to eliminate the problem on its own, signals that community norms can still play a meaningful role alongside platform moderation and legal frameworks in shaping what behavior is considered acceptable.

Read at 404 Media →
Share:X

Enjoy this story? Get the next one in your inbox.

Twice a week: the most important stories in generative image and video AI, distilled into a 2-minute read.

Free. Unsubscribe any time. No spam, ever.

Your next read

No image
Multimodal

Fine-tune video and image models at scale with NVIDIA NeMo Automodel and 🤗 Diffusers

NVIDIA and Hugging Face have joined forces to bring large-scale fine-tuning of image and video diffusion models into the NeMo Automodel framework, integrated with the Diffusers library. The collaboration aims to make distributed training more accessible for teams working with models that would otherwise be difficult to fine-tune on limited hardware. The result is a more streamlined path from a pretrained diffusion model to a customized one, without requiring deep infrastructure expertise.

No image
Multimodal

Thinking Machines Lab Releases Inkling: A 975B-Parameter Open-Weights Multimodal MoE With 41B Active Parameters And Controllable Thinking Effort

Thinking Machines Lab has released Inkling, a 975B-parameter open-weights multimodal model built on a Mixture-of-Experts architecture that keeps only 41B parameters active at any given time. Licensed under Apache 2.0, it accepts text, image, and audio inputs and offers a 1M-token context window. Rather than competing for top benchmark rankings, the model is positioned as a customizable base with adjustable reasoning depth.