Delhi Polls: Manoj Tiwari Used Deepfake to Reach Larger Voter Base

Deepfake videos were circulated in 5,800 WhatsApp groups in Delhi-NCR, according to Vice.

Ahead of the recently concluded Delhi Assembly election on 8 February, BJP President Manoj Tiwari used deepfake technology to reach out to a larger voter base.

(Photo: The Quint/Aroop Mishra)

Ahead of the recently concluded Delhi Assembly election on 8 February, BJP Delhi President Manoj Tiwari used deepfake technology to reach out to a larger vote base in which he can be heard taking a dig at the present Delhi government led by Chief Minister Arvind Kejriwal, Vice reported.

A series of videos were circulated on WhatsApp on 7 February in which Tiwari said, “Kejriwal said that he will start 500 new schools. Have they started? He said they will install 15 lakh CCTV cameras. Have they installed? He just cheated us on the basis of such promises. But now Delhi has a chance to change it all. Please press the lotus button on 8 February to form Modi-ji led government.”

While one of the videos was in English, the other one was in Haryanvi.

Vice found out that deepfake technology was used on Tiwari’s video which was originally in Hindi and not related to what the BJP president can be heard saying in the English and Haryanvi versions.

In the Hindi video, he was talking about the Citizenship Amendment Bill (CAB) being passed in the Rajya Sabha.

“Congratulations to everyone because CAB has been passed in Rajya Sabha and now for Hindu, Sikhs, Jain, Buddh, Parsis, Christians, who faced atrocities in Pakistan, Bangladesh, Afghanistan, and came to India, Modi government has provided justice to them. Now they will become citizens.”

The word deepfake comes from ‘Deeplearning’ and ‘Fake’. Deepfake technology is becoming more and more common with each passing day and given the spread of Artificial Intelligence, its getting easier and more accessible at the same time.

‘Lip-Sync’ Algorithm Used in Tiwari’s Deepfake Video

According to the Vice report, political communications firm The Ideaz Factory had partnered with the Delhi BJP IT Cell to “create positive campaigns using deepfakes to reach different linguistic voter bases.”

Speaking to Vice, Neelkant Bakshi, co-incharge, social media and IT for BJP Delhi said, “Deepfake technology has helped us scale campaign efforts like never before. The Haryanvi videos let us convincingly approach the target audience even if the candidate didn’t speak the language of the voter.”

Bakshi further informed that the deepfake videos were circulated in as many as 5,800 WhatsApp groups in Delhi-NCR and reached around 15 million people.

The Ideaz Factory claimed that they used a “lip-sync’ deepfake algorithm in which only the lip movements from the original video were altered.

“We used a ‘lip-sync’ deepfake algorithm and trained it with speeches of Manoj Tiwari to translate audio sounds into basic mouth shapes.”

Sagar Vishnoi, chief strategist, The Ideaz Factory

The political communications firm also used a dubbing artist to impersonate Tiwari and read the speech in different languages. This was then superimposed on the video.

In April 2018, American actor Jordan Peele had created a video to show how deepfakes can be deceptive. The video was doctored in a manner to show as if the former US President Barack Obama was making derogatory remarks about incumbent US President Donald Trump.

Measures Taken by Social Media Platforms

In a blog posted in January on Facebook’s official website, the company decided to remove misleading media if it fits their criteria which is mentioned in the blog, as a measure towards media that has been identified as deepfakes.

However, for videos that do not meet the criteria for removal, they can still be checked by Facebook’s partnership with independent third-party fact-checkers.

While Twitter will be putting a label on the tweets that fit into the company’s criteria of tweets containing synthetic and manipulated media.

To try and curb deepfakes, Reddit has already put out a ban. It has updated its policies to reflect that the platform will not allow “content that impersonates individuals or entities in a misleading or deceptive manner. This not only includes using a Reddit account to impersonate someone, but also encompasses things such as domains that mimic others, as well as deepfakes or other manipulated content presented to mislead, or falsely attributed to an individual or entity.”