Pangolins are the most trafficked mammal in the world, and are highly prized in Asia for their scales and meat. The African pangolin has become the focus of an AI-assisted anti-poaching initiative. (AP Photo/Firdia Lisnawati)
Effective law enforcement depends on accessing and analyzing data that can be acted on quickly. For police facing skills and funding limitations, managing vast amounts of data to generate outcomes is time consuming and expensive. But artificial intelligence (AI) is changing this. By processing massive amounts of data quickly, it can map the movements of offenders and illicit goods, identify patterns in criminal behavior and activities, and make focused connections.
AI has been recognized globally for its potential to save police hours of search and analysis work. In April, Britain’s government outlined how £230 million would be spent on AI technology to help the police save 38 million hours of police time. The European Union is implementing a project that uses AI to provide comprehensive intelligence to detect organized crime.
Policing organized environmental crime in particular is expensive, laborious and complex. These crimes often occur in remote, hard-to-access areas, involve different networks of actors, and cut across jurisdictions. Despite their significant costs to the environment, economy and society, environmental crimes tend to be a lower priority for law enforcement.
Research by the ENACT organized crime project at the Institute for Security Studies (ISS) shows how AI is doing some of the resource-heavy and complex aspects of investigating environmental crime in Africa, especially crimes related to poaching, illegal fishing, and illegal mining.
One example is TrailGuard AI, a system of cameras that enables national park officials to detect, stop and arrest poachers before they kill wildlife. The tiny cameras are easily camouflaged, and placed along trails where local intelligence has identified a threat. AI models filter out 99% of false positive images, saving battery life in remote places.
Eric Dinerstein, Director of Nightjar at the non-profit organization RESOLVE, which helped develop TrailGuard, told ISS Today that with good cell transmission, an image triggered by wildlife or poachers can reach a cellphone within around 30 seconds. This enables the appropriate authorities (e.g., park rangers or police) to mount a real-time response. TrailGuard technology was first deployed at Tanzania’s Singita Grumeti Reserve in 2018., and enabled the arrest of 30 poachers and the seizure of almost 600 kg of illegal bushmeat during a test phase in East Africa.
In 2023, Operation Pangolin was launched as a collaboration between universities, conservation initiatives, and Gabon’s National Agency for National Parks. The project collects and processes data from existing trail cameras, using AI to recognize pangolins—the most trafficked mammal in the world—from camera traps and thermal cameras. The imagery is used with Spatial Monitoring and Reporting Tool data from ranger patrols to build predictive models for pangolin poaching. The project’s long-term aim is to develop separate AI models that help predict trafficking routes and markets.
The project currently operates in Gabon and Cameroon and works closely with Nigerian stakeholders. Its team is exploring ways to build local capacity so that the data, technologies, and tools continue to be used and offer value beyond the project scope. Team member Bistra Dilkina told ISS Today that the “AI tool is an empty shell without local data. We need local champions embedded in the project.”
Other projects using AI to assist in locating environmental crimes include Skylight, a marine data platform that applies AI-powered pattern recognition, computer vision and machine learning to satellite data, uses ship movement identification and analysis from subject matter experts and rapidly applies it globally to detect illegal fishing across the oceans. It alerts coastguards and other maritime enforcement agencies to suspicious vessel patterns and locations, allowing them to assess potential non-compliant or illegal activity and distinguish it from “normal” behavior. Madagascar, Kenya, Gabon, and nations around the Gulf of Guinea are among the 70 countries that use the platform. Officials use their knowledge of national laws and their institutions’ priorities, mandates and resources, to determine how to respond. The data is received quickly enough that law enforcement can act fast, enabling timely intervention if necessary.
Digital Earth Africa (DEA) takes vast raw geospatial satellite data from across Africa and translates it into analysis-ready information. Observing changes in land use over time from satellite imagery provides insights into illegal mining activities. For instance, surface-level activities such as creating artificial ponds, clearing vegetation, and building access roads may indicate unlawful mining. Localized, real-time data on illicit mining can help make the deployment of limited resources more cost-efficient and effective. Ghana’s government has partnered with DEA to identify the location of illegal mining activities outside of mining concessions.
There are challenges and risks to using AI for law enforcement in Africa. These include limitations in the availability and volume of local data, and inadequate basic communication and digital infrastructure. A lack of technical skills and resources to respond to environmental crime, even when identified, is also an issue, as is limited investment in research and development. There are also concerns about reactive regulatory systems and data privacy, unauthorized surveillance of civilians, and criminal threats.
However, by engaging with AI’s potential, policymakers across Africa could make a real difference in the fight against organized and complex crimes. This requires a dedicated investment in building the capacity to gather large, local and relevant data sets. Budgets will also need to be allocated to digital and communication infrastructure, and generating the human capacity and skills for AI development and implementation.
African countries have an opportunity to draft and enact legislation on AI to ensure its use is regulated. The African Union’s AI white paper and roadmap could be used as a guide. Meanwhile, public-private partnerships can be leveraged to implement existing, proven AI interventions that can generate powerful crime-fighting tools.
Romi Sigsworth is a Research Consultant with ENACT at ISS. This article was published courtesy of ISS.
