An influence-for-hire operation in Angola offers a glimpse into how the battle for African elections is moving from the ballot box to the voter’s screen.
By Blessing Ivan Vava
For years, our conversations about the integrity of African elections have centred on the polling station. Could people register and vote freely? Could opposition parties campaign? Were ballot boxes secure? Was counting transparent? Increasingly, though, a large part of the election is happening somewhere you cannot walk into.
A recent investigation by the University of Toronto’s Citizen Lab into BlackCore, an Israeli influence-for-hire company, gives us a glimpse of another electoral battleground.
Citizen Lab uncovered material referring to an “Angolan Government Campaign”, involving training in storytelling, copywriting, traffic management and social-media operations. Researchers found deceptive online personas, AI-generated profile pictures, coordinated amplification and a fabricated news outlet called Agita News. Content was introduced through apparently independent identities and pushed through online communities, with its reach monitored along the way.
There are limits to what has been established. Citizen Lab could not identify the Angolan personnel involved or independently confirm the identities of those who conducted the training. After examining BlackCore’s material and related online infrastructure, however, the researchers assessed it as highly likely that the programme happened as the company described.
With Angola heading towards elections in 2027, that deserves attention. But this is not only an Angolan story. It tells us something about how African elections are changing.
I saw some of this during Zambia’s 2026 elections. I spent considerable time in the country during the pre-election, election-day and post-election periods. The rallies and polling stations mattered, but much of the political contest was happening somewhere else: on people’s phones.
WhatsApp, Facebook and TikTok had become political arenas in their own right. Fake quotations, edited videos, anonymous pages and misleading claims circulated alongside genuine political debate. In one instance, a manipulated video purported to show opposition candidate Brian Mundubile saying he had evidence that the election would be rigged.
We have seen similar tactics elsewhere. Just before Nigeria’s 2023 election, a manipulated audio recording purported to capture opposition figures discussing plans to rig the vote. During South Africa’s 2024 election, AI-generated videos circulated in which Donald Trump appeared to endorse Jacob Zuma’s MKP Party and Joe Biden appeared to threaten sanctions if the ANC won.
Some of this material was not particularly sophisticated. It did not have to be.
What makes the BlackCore revelations more troubling is the organisation behind the messaging. This goes beyond someone creating a fake photograph or forwarding a misleading WhatsApp message. Citizen Lab found evidence of fake identities, an apparently independent news platform and coordinated efforts to spread material through online communities. Generative AI has made this kind of work easier and cheaper.
Political persuasion is nothing new. Governments have communication departments. Political parties employ strategists and candidates advertise. Citizens campaign for causes and politicians they support.
The difficulty comes when we cannot tell who is speaking to us.
We know how to treat a message carrying the logo of a political party. We can also recognise official government communication. It becomes harder when the same political message appears to come from an ordinary citizen, an independent commentator or a news organisation that is not what it claims to be.
This connects with something I have been writing about for some time: Africa’s digital sovereignty.
Much of that debate has focused on infrastructure, data centres, foreign technology companies and where our data eventually ends up. I have argued that these are political questions as much as technological ones. Elections bring another part of the problem into view. The issue is no longer only who controls our data or the infrastructure underneath our digital lives. It is also about the growing power to influence what reaches our screens in the first place.
That power is easy to underestimate. Most people scrolling through Facebook, TikTok or WhatsApp are not thinking about the systems behind what they are seeing. Yet a network of fake accounts can make an argument appear more popular than it is. Repetition can give a fabricated story the appearance of credibility. Algorithms then determine how far some of this material travels.
In an election, attention itself becomes part of the contest. Winning it can mean deciding which issue dominates the day, which allegation refuses to disappear and which political narrative follows voters around on their phones. That is why digital sovereignty must also concern itself with the conditions under which our attention is captured and political opinion is formed.
This creates a problem for election observation.
Our methods were largely developed for a different kind of election. We know how to observe voter registration, rallies, polling stations, counting centres and results forms. An observer can arrive at a polling station in the morning and confirm that the ballot box is empty.
It is much harder to observe everything that happened on the voter’s phone before they arrived there.
That does not mean every misleading post changes a vote, nor should we start blaming technology whenever an election produces a result we do not expect. Voters have agency. Misinformation is also much older than social media. South Africa’s experience showed that an old photograph with a false caption can sometimes do the job just as effectively as an AI-generated video.
What has changed is how quickly this material can be produced and spread, and how difficult it can be to know who is behind it.
Election observation will have to catch up. Observer missions and civil society organisations increasingly need people who understand digital platforms, coordinated networks, AI-generated material and online political advertising. Election management bodies will also have to engage platforms and independent researchers more seriously.
There is a danger here too. Governments can easily use the language of fighting “fake news” to restrict journalism, opposition politics and legitimate criticism. Protecting elections from manipulation cannot become an excuse for governments to decide what citizens are allowed to say or read.
That is why what happened in Angola should concern the rest of the continent.
The important lesson from BlackCore is not simply that false information exists online. We have known that for years. More disturbing is that the tools for creating the appearance of public opinion can now be packaged, outsourced, and sold.
That changes the electoral integrity conversation.
The ballot box still matters. But before a voter reaches it, another political contest may already have been taking place for weeks on the small screen in their hand.
If we watch the ballot but ignore that contest, we may increasingly be observing only half the election.ENDS
About the Author
Dr Blessing Vava is a Zimbabwean researcher, writer and democracy advocate working on elections, digital sovereignty and governance in Southern Africa. He holds a Doctor of Literature and Philosophy (DLitt et Phil) in Communication Studies from the University of Johannesburg. His research examines the intersection of technology, geopolitics and governance in Africa. He writes The ThursdayBriefing, a regular commentary on democracy, technology and politics.
No comments:
Post a Comment