In the weeks before Indonesia’s February 2024 election, a deepfake video of former dictator Suharto, who died in 2008, spread on social media. As recorded in the OECD’s AI Incidents Monitor, the clip was tied to the Golkar party and showed an AI-generated Suharto urging voters to back its candidates, his face and voice synthesized to deliver a political appeal he never made. The OECD logged it as a realized AI incident rather than a hypothetical risk, citing manipulation of the electorate and the undermining of democratic processes.
The Suharto video was the darker edge of a campaign already saturated with generative AI. The eventual winner, Prabowo Subianto, a former general from the Suharto era, was rebranded through an AI-generated “gemoy” (cuddly) cartoon avatar that flooded TikTok and merchandise, softening a strongman image into something approachable. Against that backdrop, reanimating the actual Suharto to endorse candidates showed how the same technology could move from image-making to outright historical fabrication.
For a business reader, the case is a clear example of synthetic media used to put words in the mouth of someone who cannot object, and of how quickly AI can blur the line between flattering rebrand and deceptive impersonation in a high-stakes public contest.