South Korea to Invest $6.2 Million in AI System to Detect Deepfakes and Misinformation

SEOUL — South Korean police will invest 9.1 billion won ($6.2 million) to develop a system capable of detecting deepfakes and misinformation, as the country grapples with a surge in AI-generated crimes, The Korea Times reported Sunday.

The system, expected to be completed by December 2027, will use a multimodal algorithm to analyze noise and sound frequency, helping authorities identify deepfake videos and AI-generated voices, officials said. The initiative also includes leveraging generative AI to track the origins of online echo chambers in an effort to curb fake news before it spreads.

Authorities anticipate that the new system will significantly reduce the time required to detect manipulated media. The urgency comes as police reported 573 arrests linked to 1,094 deepfake-related sex crimes in the first 10 months of 2024—many involving minors.