An Indian election candidate created multiple deepfakes of his own video with him speaking different languages to access voters, reaching ~15M WhatsApp users
The 2020 VICE story is the early data point in an arc the related coverage traces directly: WhatsApp was already flagged in 2018 as a misinformation vector in Indian elections, and this candidate's self-made multilingual deepfakes — reaching roughly 15 million users — showed synthetic video entering that same channel, this time from campaigns themselves rather than rumor networks.
By the 2024 cycle the practice had industrialized: politicians moved to voice-cloning tools for personalized voter messages in late 2023, and by election time an estimated 50M+ AI voice calls were placed in two months. The candidate's ~15M-user experiment reads as the proof of concept that scaled.
First-order effects
The candidate gains access to language-diverse voter blocs a single recorded video could never reach, at near-zero marginal cost per language version — a direct campaign-capability advantage over rivals still producing one video per language.
WhatsApp becomes a confirmed distribution channel for campaign-produced synthetic media, not just forwarded rumors, putting the platform's moderation posture on the spot.
Second-order effects
Competing campaigns are pushed to match the tactic — the 2024 coverage of deceased-politician deepfakes and mass AI voice calls shows rivals adopting the same tooling once one side demonstrates reach.
Cheap voice- and video-cloning vendors gain political parties as a customer class, pulling campaign budgets toward AI tooling and away from traditional production.
Third-order effects
If the pattern holds, multilingual synthetic outreach becomes table stakes in large multilingual democracies, shifting campaign competition from message creation to distribution scale — and concentrating leverage with the platforms that gate that distribution.
Election authorities face structural pressure to define disclosure and authenticity rules for campaign-generated AI media, a gap the 2018–2024 coverage shows widening with no regulatory layer in place.
The trend: Indian political campaigning is shifting from organic WhatsApp forwarding to industrially produced synthetic media, with one candidate's ~15M-user deepfake experiment in 2020 scaling to 50M+ AI voice calls by the 2024 elections.
In a possible first, a political candidate in India used deepfake technology to generate and distribute fake videos of _himself_ https://www.vice.com/... by @NilChristopher He did this to pander to different audiences in languages he didn't speak.
“Tiwari's fabricated video was used widely to dissuade the large Haryanvi-speaking migrant worker population in Delhi from voting for the rival political party”. Bakshi says deepfakes on 5,800 WhatsApp groups in Delhi and NCR reaching approx 15 mill ppl. https://www.vice.com/...
“It's 2020, and deepfakes have become a powerful and concerning, tool that allows humans to manipulate or fabricate visual and audio content on the internet to make it seem very real.” https://www.vice.com/...
“We define “AI-Mediated Communication” (AI-MC) as communication between people in which a computational agent operates on behalf of a communicator by modifying, augmenting, or generating messages to accomplish communication or interpersonal goals” https://twitter.com/...
Most discussion of potential deepfakes use in campaigns has focused on how they could be used to target opponents. But this shows how they can be used - in combo with microtargeting - to help candidates directly manipulate voters about themselves. https://twitter.com/...
🤔 This takes me back to a convo I had at State where we said “wouldn't it be cool if we could use technology to have the Secretary speak other languages?” and then the ethical conversation that ensued. Well, hope they solved it after I left cuz here we are. https://twitter.com/..…