Yoon Suk-yeol presidential deepfake candidacy prompts concerns
Occurred: February 2022
Report incident ๐ฅ | Improve page ๐ | Access database ๐ข
A campaign by South Korean politician Yoon Suk-yeol to become the world's first 'deepfake candidate' to win a presidential election raised concerns that it painted an inaccurate picture of his real character.
'AI Yoon''s team turned to avatars and short-form videos to explain policy ideas and lampoon his rival in a bid to win over younger voters.
However, some voters have complained that it is an inaccurate physical representation and paints an untrue political picture of Yoon. Others have pointed out the potential for misuse of deepfakes in the form of mis- and disinformation.
AFP says South Korea's election monitor allowed AI candidates to campaign on the condition deepfake technology is clearly identified, and does not spread misinformation.
A recent study published in the Proceedings of the National Academy of Sciences USA suggests that real humans can easily fall for deepfake faces and tend to interpret them as more trustworthy than the real thing.
System ๐ค
Operator: People Power Party
Developer: People Power Party; DeepBrain AI
Country: S Korea
Sector: Politics
Purpose: Communicate with young voters
Technology: Deepfake - audio, video; Generative adversarial network (GAN); Neural network; Deep learning; Machine learning
Issue: Accuracy/reliability; Mis/disinformation; Ethics
Transparency: Governance
Research, advocacy ๐งฎ
Nightingale S., Farid H. (2021). AI-synthesized faces are indistinguishable from real faces and more trustworthy
Page info
Type: Incident
Published: March 2022