By A Mutambara
Folks, the video being circulated as a beautiful and inspiring speech by the incoming President of Senegal has been proven to be a deepfake!!
A deepfake is synthetic media generated using AI techniques, particularly deep learning algorithms.
It involves manipulating audio, images, or videos to create content that appears real but is actually fabricated or altered.
The deepfake in question was first released by an X (ex-Twitter) user,
@von_Bismack, on Tuesday, 26 March 2024.
It was subsequently reposted on many social media platforms such as Facebook, WhatsApp, Instagram and TikTok
It went viral.
What are the facts?
The person in the video is not the winning Senegalese presidential candidate, Bassirou Faye, in the elections held on Sunday, March 24, 2024.
It is his mentor and party—Patriots of Senegal (PASTEF)—leader Ousmane Sonko (who was barred from contesting the election because of a politically contrived conviction).
Sonko backed Faye in the polls.
Of course, correctly identifying the man in the video was the easy part.
The viral video is a manipulated version of a press conference held by Sonko in 2021. Keyframes in the viral video (the deepfake) were subjected to reverse image search, and it was discerned that the visuals of the clip were extracted from the speech delivered by Ousmane Sonko at a press conference in July 2021 in Dakar, Senegal.
He was speaking on an entirely different subject: Against the proposed decision of the Senegalese government to adopt an anti-terrorism law.
More significantly, Sonko addressed his audience in French, not English, as seen in the viral video. He did not utter a single word of English in that presser.
AI technology was used to alter Sonko’s voice and lips in the clip to give the impression that he spoke in English and to change the content of the speech.
What are the dangers of such deepfakes?
1) It is prudent to note that it took three days for the Senegalese deepfake to be unravelled and exposed.
Still, not many people have received the information that it was fake, and some will never get the memo.
2) In this case, hopefully, not much damage was done. The incoming President is a young, dynamic, progressive leader.
A bright future is on the horizon for Senegal and Africa.
3) However, consider a hypothetical situation where such a deepfake is released (and goes viral) against a company or product, and it takes three days for the deepfake to be debunked.
Can you imagine the damage done to the business: decisions taken against it, falling share price, loss of customers, brand damage, etc?
Can all these be reversed after the deepfake is debunked after three days? Food for thought.
4) Another hypothetical situation: Suppose there is a close election, and two days before the polls, such a deepfake is released against a party or candidate.
The deepfake goes viral and is then only unravelled three days later, a day after the elections.
Can the impact of the deepfake on the elections be reversed? Food for thought
Welcome to the brave new world of AI.
Indeed, there are great opportunities, but there are also potential dangers and risks.
