How Is Artificial Intelligence Redrawing the Map of Conflict in Africa?

In late 2021, Tigrayan forces were advancing toward Addis Ababa. Armed drones bought from several foreign suppliers helped stop them, and within a year Ethiopia's government was at a negotiating table it had been losing its way toward. Four years later, in the Sahel, the same category of technology — cheaper, commercially available, and increasingly autonomous — is being flown by insurgents against the armies of Mali and Burkina Faso.

That reversal is the whole problem in miniature. Artificial intelligence is neither a peace technology nor a war technology. It is an amplifier. What it amplifies depends almost entirely on the institutions that hold it — and across a continent marked by fragile states, thin infrastructure, and low public trust, that is not a reassuring conclusion.

From holding ground to holding the data

AI has moved the center of gravity in African conflicts. Power used to be measured in territory held, fighters mobilized, and foreign money secured. Increasingly it is measured in information superiority — the ability to gather battlefield data, process it quickly, and act on it before an opponent can.

This is less abstract than it sounds. Target-recognition software and the fusion of satellite imagery, sensor feeds, and intercepted communications have compressed what militaries call the sensor-to-shooter cycle — the gap between spotting something and striking it — from hours to minutes. In open, weakly governed terrain with little natural cover, that compression is decisive.

It is also dangerous for civilians. The same systems that sort convoys from farm traffic also sort people: into categories, watchlists, and targets. Information superiority and mass surveillance are the same capability pointed in different directions.

Three battlefields

Three theatres show how far this has already gone.

Ethiopia. Between 2021 and 2022, drone fleets assembled from multiple suppliers — Turkish, Iranian, Chinese, and Emirati sources have been documented by open-source researchers and satellite-imagery analysts — let the federal government strike Tigrayan positions and supply lines at low cost. They halted the advance on the capital and helped push the war toward the settlement reached in Pretoria in November 2022. But drones could not clear insurgents from mountainous terrain, could not touch the political and ethnic grievances driving the war, and deepened Ethiopia's dependence on outside arms suppliers. What they produced was a longer war and a rearranged set of foreign alliances — not a decision.

Sudan. Since 2023, the war between the Sudanese Armed Forces and the Rapid Support Forces has become a testing ground. Drones, satellite-imagery analysis, digital monitoring, and open-source tracking now sit alongside conventional combat. The trajectory has only steepened: the Rapid Support Forces have moved from improvised munitions to long-range strike platforms, while the army has fielded Iranian and Turkish systems. Semi-autonomous drones extend reach while reducing the need to close with the enemy — which also raises the risk of error and the humanitarian cost. Sudan is where the internationalization of an African war runs most visibly through supply chains.

The Sahel. Here the direction of travel reverses. Jama'at Nusrat al-Islam wal-Muslimin and Islamic State affiliates have adapted commercial drones for reconnaissance, propaganda, and dropping improvised explosives on isolated military positions. Analysts date the group's first armed drone strike to September 2023; attacks are now recorded across Mali, Burkina Faso, and Togo. The tactical consequence is concrete: earthworks no longer protect a garrison, and Sahelian forces now need radio-frequency jamming and signal detection they largely do not have. Non-state actors have converted civilian technology into military capability without building conventional armies to do it.

Against this, early efforts at localization matter more than they might appear. Nigeria has moved into domestic drone production; South Africa builds platforms designed for coordinated multi-unit operation. Both point toward strategic autonomy rather than deeper dependence.

The mirror image

The peacebuilding side of the ledger uses the same capabilities, and this is where the ambivalence becomes unavoidable.

Early-warning systems are the clearest win. Machine-learning models that combine rainfall, crop failure, and water-scarcity data with political and demographic indicators can flag violence risk months ahead — the Water, Peace and Security partnership's global tool, applied in Mali's Inner Niger Delta, reports correctly predicting 86 per cent of conflicts in testing. The limitation is institutional, not technical: a forecast is worthless without the capacity, mandate, and accountability to act on it.

Natural language processing can help mediators map competing narratives and locate overlapping interests. AI can flag coordinated disinformation and hate speech in near-real time — a capability that matters in Nigeria's counter-extremism efforts against Boko Haram and Islamic State West Africa Province, where the value lies less in deleting posts than in mapping which accounts actually drive radicalization. Machine analysis of images, video, and testimony can build evidentiary records for transitional justice and guard collective memory against denial.

Each comes with the same structural warning. Algorithms trained on the wrong data misread local languages, suppress legitimate speech, and reproduce bias. And none of them supplies the trust, cultural judgment, or political legitimacy that a durable settlement requires.

Why the promise stalls

The constraints are severe. Electricity shortages and patchy connectivity in places like Mali and the Central African Republic undercut the very early-warning systems meant to serve them, while leaving governments and UN missions exposed to cyberattack. Reliable data on populations, armed groups, and violence is scarce, and where states depend on external actors to collect it, the resulting picture can carry someone else's priorities — inflating some threats, ignoring others.

Capability gaps compound this. Nigeria's technological base far exceeds that of Chad or Niger, which means AI-assisted security produces uneven results across borders that armed groups cross freely. Donor dependence leaves projects tethered to foreign timelines: in Somalia, most digital-security initiatives have run on external support without ever being absorbed into state institutions. And in countries with long records of surveillance and repression, deploying these tools without clear ethical rules deepens public suspicion rather than building confidence.

The uncomfortable conclusion

AI will not start a war in Africa, and it will not end one. What it will do is magnify whatever governance surrounds it — which is why the same handful of capabilities appears on both sides of the ledger, once as a peace opportunity and once as a conflict risk.

What follows from that is a policy agenda, not a technology agenda: control over data and digital infrastructure as a question of sovereignty rather than procurement; regulation of military and surveillance applications before capabilities outrun oversight; regional cooperation to close capability gaps that armed groups already exploit; and a coordinated African position in the international forums where the rules for artificial intelligence are being written by others.

Technology has never resolved a political grievance. What it can do is decide how quickly the grievance turns into casualties.

Download the full study here.

Empowerment

Amplifying African voices for sustainable progress together.

Contact US

Growth

• Street No. 3281, N'Djamena, Republic of Chad.
  

• info@africannarratives.org

© 2024. All rights reserved.