An artificial intelligence system operated by the Center for Digital Sexual Crime Victims in South Korea flagged an online post as a likely child sexual exploitation risk with an 86% probability on November 10, 2026, catching solicitations before actual harm occurred.
- The Center for Digital Sexual Crime Victims deployed the AI detection system in April 2026 to automatically monitor social media and online platforms 24 hours a day.
- During a press tour demonstration, the technology scanned 14 platforms using roughly 220 core keywords to analyze text and image data simultaneously.
- Staff intervention occurs when the AI classifies a post with a 70% overall exploitation probability and a 60% child sexual exploitation probability, leading to removal requests.
Employees previously reviewed thousands of daily posts manually, creating severe bottlenecks outside standard working hours.
The AI tool now processes roughly 5,000 posts every single day. Individual staff processing rates surged from an average of 20.8 items per day last year to 202.1 items per day between April and September of this year, according to mt.co.kr figures.
Center officials plan to scale monitoring operations from 14 platforms up to 139 platforms starting next year. The upgraded architecture will move beyond static keyword searches to analyze conversational flow and context directly.
“A teenager posts a short video, and comments like ‘pretty’ or ‘how old are you’ start building trust, which is how online grooming begins,” Center Director Kim Misu explained during the briefing, noting that simple keyword searches often fail to catch these grooming patterns.
