South Korea’s Gender Equality and Family Ministry said on 10 November that an AI system introduced by the National Center for Digital Sexual Crime Response in April had identified 96,210 items of online child sexual abuse material between April and September. The centre notified platforms to remove the material, Anadolu reported. The ministry said the system had increased daily detections nearly tenfold. Its announcement concerned a centre that also helps victims seek removal of illegal material and pursue legal action.
Key points
- The centre identified 96,210 items between April and September and notified platforms to remove them.
- Workers review material flagged by the round-the-clock system before filing removal requests.
- Daily detections averaged 202.1 per monitoring worker, against 20.8 last year, according to Anadolu.
- The ministry plans to monitor 139 platforms next year. Reports put this year’s coverage at either 74 or 79.
Centre staff review AI flags before removal requests
The system runs around the clock across online services including social media and random chat applications. It flags suspected illegal material for centre workers, who review it before submitting a removal request to a platform. Staff had previously monitored the services themselves, The Korea Times reported. Human review therefore remains between an automated flag and the request sent to a platform. The reported 96,210 items are material the centre identified and notified platforms to remove, rather than a count described as completed removals.
The centre was established in 2018 to support people affected by digital sex crimes. Its work includes seeking removal of illegal online material and taking legal action. Last year it supported 10,637 people, among them 3,052 victims under 20. The AI system gives its staff another way to find material for review across the services it monitors, alongside the centre’s work with victims. The ministry’s figures cover the six months from April to September, beginning when the centre introduced the system.
The 202.1 daily detections measure
Daily detections averaged 202.1 per monitoring worker, compared with 20.8 last year, Anadolu reported. The Korea Times described the same figures as daily averages of identified material without a per-worker qualification. Both accounts attribute the figures to the ministry. The comparison places the new system’s reported detection rate against last year’s work, when centre employees monitored platforms directly. It measures detections, while the reported process still requires a worker to examine flagged material before a platform receives a request.
South Korean police detained 3,557 suspects in online sexual abuse cases between November 2024 and October 2025. Teenagers accounted for nearly half of the suspects, and deepfake-related offences represented 35.2% of cases. Those police figures cover suspects and offences across online sexual abuse cases. The centre’s six-month figure counts material it identified through the AI system and notified platforms to remove. The two measures concern different work by different bodies.
Other organisations are also applying AI to a narrower part of the online distribution problem: AI Affairs previously reported on Meta’s tools for detecting ads leading to child sexual abuse material. South Korea’s system monitors platforms for the material itself, then passes suspected items to centre staff. In the ministry’s reported process, a flag begins a review that can lead to a removal request.
Ministry plans to monitor 139 platforms
The accounts differ on the number of platforms covered this year. Anadolu put it at 74. The Korea Times put it at 79. Both reported the ministry’s plan to extend monitoring to 139 platforms next year.
The ministry also plans to improve the system’s accuracy by training it with the centre’s database, The Korea Times reported. That work sits alongside the proposed increase in platform coverage. Centre staff review flagged material before filing removal requests.
