By: Peter Karanja, Autonomy Global Ambassador – Africa
African conservation does not have a drone adoption problem. Over the years, drones have been deployed to track wildlife. AI has been used to identify animals from drone footage. And tools exist that connect ranger patrols, sensors and control rooms with decision makers. The more important question is not whether tools exist or even whether people use them. The key issue now revolves around whether they are being used in ways that actually solve problems in conservation. Discussions at the March 2026 Global Conservation Tech and Drone Forum in Kenya tackled this issue head on. This article captures some of the event’s major themes.
Drones As Critical Conservation Tools
For a long time, drones were treated as a novelty. The conservation sector has moved well past that. Drones are no longer hard to acquire. People have seen how well they work. Now they are interested in how hardware capabilities can be extended through software.
For example, AI is now more accessible to use and customize to specific use cases. With AI, drones can see wildlife from above. Software can now turn raw aerial footage into behavioral data at a level that was previously difficult to collect. As one example, research presented at the event described using drone imagery and computer vision to identify animals frame by frame and then connect those detections into movement trajectories. Even so, what does that mean? One speaker captured the issue, when they said, “all this data needs to be analyzed somehow,” because data on its own “is not very helpful.”
Another speaker, when discussing drone use cases, noted that we should not just be concerned about where drones are being used, but also about where they are not being used, and why. In many African conservation contexts, the barrier to drone use is not due to a lack of awareness or even hardware, but the operational difficulty of running a sustained program.
The Operational Design Problem
As drones and AI become more commonplace, the challenge seems to involve an operational gap. Historically, conservation-focused drone programs launched after an incident instead of in a proactive manner. Proactive programs require battery readiness, trained crews, command visibility and a response process that starts before the flight and continues after landing.
Yet progress has been made. Compared to a few years ago, conservationists now want to run in-house programs, instead of hiring an outside drone company. They want their own rangers to fly the drones and handle the various technologies. And this makes sense since drones and any other new tech works best when embedded with the current workflow.
In-house capability, however, raises the standard. Once an organization creates an internal drone program, then it must function as a real operational unit. That means maintenance, spare batteries, standard operating procedures, scheduling, pilot continuity, reporting and data management. It also means leadership must understand the purpose of the drone team. Will they conduct surveillance, wildlife counts, human-wildlife conflict response, habitat monitoring or all of the above?
As one speaker noted, the future of drone operations for the sector will require continuity, knowledge transfer and educating communities. This must be part of the conversation, alongside funding for technology itself. That matters because too many discussions on autonomy still assume that technology scales mainly by becoming more advanced. In conservation, it often scales by becoming more usable and maintainable to the people running it. Some drone deployments remain reactive simply because the system surrounding the aircraft has not matured enough for persistent monitoring.
Integration More Important Than Flight
Tools don’t matter if they work in isolation. A drone by itself produces footage. A drone connected to patrol data, sensors, mapping layers and command-room processes produces operational intelligence. One panelist emphasized this last point. They described the value of being able to pull a livestream of an incident to the command center because it allows them to decide whether to send rangers or choose another method of response.
The same systems thinking came through in a session focused on scaling technology. The discussion emphasized that meaningful progress requires collaboration among conservation, drone engineering and AI. In other words, the next phase of conservation technology in Africa will likely be won by whoever can connect aircraft, software, people and decisions into one functioning workflow.
This especially applies to AI, which panelists agreed is only useful if it can be integrated fully into a given system. AI eases the burden of manually sorting through the columns of data that drones, camera traps, and other tools produce. In conservation, AI provides the most value when it sits between raw data and a human decision-maker for tasks like detecting animals, filtering imagery, flagging anomalies or simply reducing time spent on repetitive reviews.
Even so, while AI may be effective, it should not be used to replace field expertise. Instead, all tools should “augment and amplify human expertise.” And that expertise, both technical and managerial, remains key. AI cannot solve a poorly designed operational chain. If the team does not know who receives the alert, who validates it, who launches or what action follows, the AI output simply may join the pile of unused information. This is why the future of conservation autonomy in Africa will depend on whether organizations can integrate AI into working routines.
People Remain the Core Infrastructure
Speakers also emphasized the irreplaceable role of rangers and local teams. When new technology arrives in a conservation landscape, local communities are typically the last to be informed, and the last to be credited. This happens even though their institutional knowledge and long-term relationships with the land are often the reason any program functions at all.
Local communities, frontline practitioners and on-the-ground researchers must be formally recognized as legitimate stakeholders in conservation data itself. Many of the datasets powering today’s AI models exist precisely because those communities allowed collection in the first place and because practitioners gave that raw data meaning.
Conservation technology is never deployed into a vacuum. It enters landscapes already defined by ranger expertise, community trust, local risk dynamics and deeply embedded institutional habits. Organizations that treat the human layer as secondary, or assume it can be engineered around, will find their most sophisticated tools will become a system that can detect a threat but cannot reliably act on it.
Data Governance As A Systems Problem
The forum made clear that the conversation doesn’t end with tools or even operational design. It ends, and increasingly begins, with data governance. Collaboration without defined rules can quietly become extraction, particularly where power imbalances already exist with conservation data in Africa. As drone programs expand and digital monitoring platforms multiply, a critical question must answer who controls access, who sets terms and who ultimately captures value from what is gathered.
Consider the trajectory of platforms like EarthRanger, the U.S.-developed conservation software now deployed across more than 130 protected areas in 34 countries, the majority of them in Africa. The platform aggregates ranger patrol data, drone feeds, sensor inputs, and wildlife tracking into a single real-time operational picture, and it does it well. But the intelligence it produces, the movement patterns of elephants, the patrol routes of rangers, the poaching hotspot maps built from years of field observation, lives on infrastructure developed, hosted and governed externally. African park managers gain powerful situational awareness. What they often don’t control is the data architecture beneath it, the terms under which that data is stored, shared or eventually monetized.
WildDrone, an initiative working with universities to develop wildlife detection models, serves as another example. PhD students may collect drone imagery in African protected areas as part of academic research, then use that data to train AI systems that could later become commercially valuable products. This can create difficult questions. Who owns the original dataset: the student, the university, the park authority, the community where the data was collected or the company commercializing the model? Who benefits if the final product is sold globally?
This is not an indictment of any platform. It is a structural problem. As drone programs expand and digital monitoring systems multiply, a critical question must be answered: who controls access, who sets terms and who captures value from what is gathered? If African conservation organizations develop increasingly sophisticated monitoring infrastructure while remaining dependent on external platforms, external analytical frameworks, and externally defined data protocols, the continent risks supplying the raw intelligence while others determine its downstream worth. That is not only a question of sovereignty, but also a direct operational risk. Systems perform best when the people closest to the mission understand, trust, and have genuine authority over the information those systems produce.
What the Sector Needs Next
African conservation surely has sufficient enthusiasm around drones. What it needs is more discipline around operations. That means redirecting serious attention, and serious funding, from isolated tool deployments toward the less visible but more consequential work of building durable systems (maintenance protocols, internal pilot capability, data review pipelines, command-room integration, training continuity and governance structures) that keep technology accountable to conservation priorities rather than donor timelines or vendor roadmaps. The organizations that lead in conservation technology on the continent will be the ones who have internalized the hard truth that drones, data and AI only become genuinely valuable when embedded in a system capable of interpreting them, acting on them and sustaining that capacity over time.