Your Africa security plan was written for the last war
Cheap drones and ordinary chatbots are making small armed groups more capable where South African companies operate.

Sign up for my Substack daily AI newsletter https://johanosteyn.substack.com
Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/
When the United Nations warns about AI in warfare, most boards picture great powers and autonomous weapons, a debate for diplomats. In October, the director of the UN Institute for Disarmament Research added a warning closer to home: as AI tools become easier to use, it will become easier for non-State actors to carry out ever more sophisticated military operations. In parts of Africa where South African companies run mines, towers, stores and supply routes, that is already happening.
CONTEXT AND BACKGROUND
The scale has changed quickly. The 2026 Global Peace Index records that drone strike events rose 115-fold between 2018 and 2025, and that 565 different armed groups carried out at least one drone attack in that period.
In the Sahel, armed groups have moved from dropping explosives off commercial quadcopters in 2023 to drones guided by fibre-optic cable, which jamming cannot stop, and coordinated drone and ground assaults on airfields. Four groups now operate armed drones across Mali, Burkina Faso, Niger and Nigeria, and one research paper documents a claimed swarm of about 25 drones launched against Gao airport in Mali in March 2026. Africa Defense Forum, a magazine published by US Africa Command, reports that some recent attacks used AI-enabled drones able to locate and strike targets without a human operator, and quotes the analyst Niccola Milnes: "You use AI to teach you how to use AI".
INSIGHT AND ANALYSIS
The second change is quieter. A University of Cambridge study, based on 57 interviews with 27 former members of ISWAP and Boko Haram in north-eastern Nigeria, found that both factions used ordinary AI chatbots for weapons troubleshooting, explosive device design, tactical planning and logistics. Both set up dedicated AI units, and members bypassed safeguards through jailbreaking and multiple accounts. The author is careful to note that the study cannot conclusively establish real uplift, and that the uses documented remain conventional.
I have previously written about this in the context of cybercrime, arguing that chatbots have removed the skill barrier that once limited serious attacks. The same collapse is now visible on the ground.
The costs are real. In Sudan, the UN reported that more than 1,100 civilians died in drone strikes between January and June 2026, about 80% of recorded civilian deaths in that period.
The fair objections deserve weight. Most deaths in African conflicts are still caused by conventional weapons, and much of what is called AI warfare is cheap hardware rather than clever software. Security is the duty of states, not company boards. Governments are also using AI and drones for early warning and border protection. All of this is true. None of it changes the fact that the threat a mine or depot faces has gained a dimension that most security plans never considered.
IMPLICATIONS
A threat in the Sahel or Sudan matters in Johannesburg because group risk, insurance and supply chains are usually managed centrally, across every country of operation. For boards of companies operating in conflict-affected regions, three questions follow. First, does our security assessment assume an attacker who can see from the air and strike from a distance, and when was it last updated? Second, do our insurance and political violence cover respond to drone attacks, and do our duty-of-care and evacuation plans reflect them? Third, for companies that build, sell or deploy drones or AI tools in the region, what do we know about where they end up? These are not questions for the security manager alone. They belong on the risk committee's agenda, alongside currency and political risk.
That last question is getting an international answer. UNIDIR is developing a voluntary framework of responsible industry behaviour for military AI with the UN human rights office, and argues that companies make decisive choices about data, testing and autonomy years before a commander uses a system. African companies should help shape it rather than inherit it.
CLOSING TAKEAWAY
The military AI debate has focused on whether machines will be allowed to decide who lives and dies. That question matters. On the continent where many South African companies make their money, a more immediate change is under way: cheap drones and ordinary chatbots are giving small armed groups capabilities once reserved for armies. Boards that last reviewed their security plans before that happened are protecting their people against a threat that no longer exists.
Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence's impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan's work at https://www.aiforbusiness.net




Comments