South KoreaPublic sector2024 – ongoingKorean National Police Agency
Deepfake detection with forensic review
Korean police used deepfake detection software in 1,636 video examinations
Problem
Investigators needed to establish whether suspected deepfake videos had been manipulated. Some recordings remained ambiguous after automated screening.
Approach
Police deployed software that examined facial manipulation, biological signals and file metadata. They continued training its AI models for Korean deepfake material. Forensic video analysts reviewed ambiguous results, while digital specialists analyzed the AI tools used to produce manipulated recordings.
Results
Takeaway
Automated screening fed a forensic review process. The usage count did not establish detection accuracy or the software's contribution to arrests.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Partner co-development
- Evidence
- Company's own claim
Sources
Source summaries
Police reported using deepfake detection software in 1,636 video examinations since its development. Ambiguous cases went to forensic video analysts, supported by analysis of AI tools used in the offenses.
Full details are in the original sources.