Elections in Africa: AI generated deepfakes could be the greatest digital threat in 2020

Abdou Razak (C) of Togo demonstrates with others against President Faure Gnassingbé

By: Babatunde Okunoye 

The year 2020 will be a year of many elections in Africa. At least 21 nations – including Ethiopia, Togo, Sudan, Mali, Gabon, Egypt, Chad and Cameroon will conduct Presidential or Parliamentary elections in the course of the year.

In the past 5 years, digital rights violations such as Internet disruptions and arrests of bloggers and Journalists have often coincided with elections in Africa. In the cut-throat political brinkmanship Africa has been known for, elections are high stakes events where digital rights are routinely relegated to achieve political ends. The eight nations mentioned above in particular have a history of Internet disruptions during political events. As we approach the 2020 election cycle therefore, digital rights advocates are on alert to the possibility of Internet disruptions and other digital rights violations.

Nevertheless, the rapid development of Artificial Intelligence (AI) and disinformation techniques globally raises the possibility of a new threat to stability during the 2020 election cycle in Africa. This new threat comes in the shape of Artificial Intelligence generated deepfakes. Deepfakes are AI-generated hyper-real but fake videos purporting to show particular individuals saying or doing whatever the generator of the video intends to display. Deepfakes have already attained notoriety in the United States. For instance, a deepfake video featuring the likeness of Facebook CEO Mark Zuckerberg stating “whoever controls the data, controls the future” surfaced on the Internet on the eve of United States Congressional hearings on Artificial Intelligence he was scheduled to attend last year.

Similarly, researchers at the University of Washington, USA had circulated online a deepfake video of President Obama in which they made him say whatever they wanted. Also, a doctored video of Nancy Pelosi, the Speaker of the United States House of Representatives appeared online, giving the appearance of her being drunk and slurring her words. Although there was no evidence that this particular video was AI-generated, its effect could be gauged by the fact that it was shared on Twitter by Donald Trump, the United States President.

In Africa also, during his traditional new year presidential address of 2019 in Gabon, President Ali Bongo was reported to look off during the video broadcast, which triggered suspicions in some quarters that this video broadcast was an AI-generated deepfake of the President. The uncertainty created by this scenario within the context of the President’s illness led to an aborted coup attempt and Internet disruption in the central African country. There was no clear-cut evidence that this particular video was a deepfake, but the very existence of the technology within the fragile political context existent in Gabon at the time created uncertainty.

Africa, much like other regions of the world, has already demonstrated its vulnerability to misinformation campaigns through the Cambridge Analytica scandal which had footprints in Kenya and Nigeria. Given the high levels of illiteracy on the continent, it is not far-fetched to imagine how a well-coordinated AI deepfake campaign could stoke social unrest around elections, particularly in highly contested contexts like Gabon. Imagine a deepfake video showing a prominent leader making inflammatory comments towards another ethnic or religious group, or of election commissioners reading fake election results which then go viral and are widely accepted as original before being debunked. Africa has a long and illustrious history of violence and carnage around elections, and scenarios such as these can well be tinder for the fire. As already clearly stated, the very presence of AI deepfake technology sows doubt on the originality of even real video broadcasts, a situation which further spawns uncertainty and instability.

The year 2020 represents another opportunity for digital rights advocates to be on the watch in defence of digital rights. But not just digital rights as we’ve traditionally known them. As the digital realm expands to spawn more technological advances like Artificial Intelligence, advocates need to keep up with these advances and be aware of the human rights and socio-political consequences of new technologies. They cannot afford not to, for much is at stake, not least the safety and stability of the nations they call home.

Babatunde is Research Officer at Paradigm Initiative. Email:  babatunde.okunoye@paradigmhq.org