tackling wildlife cybercrime in the EU

Law enforcement agencies (LEAs) and nongovernmental organisations (NGOs) still rely largely on manually collecting data from the internet for identifying wildlife cybercrime, a very labour and time-intensive exercise. Technology is often regarded as a potential game-changer in this area. There is an untapped potential for automated data analytical tools to provide relevant information. Beyond accelerating and improving their monitoring activities, these tools could help practitioners to draw up a more comprehensive picture of the true scale and nature of wildlife cybercrime in the EU, including identification of where it occurs on the internet and which species are most affected. Developing this picture is important for supporting investigations, for governments to develop a sense of how best to respond, and for effective engagement with the online technology companies whose platforms are unwittingly being used.

This report offers a general assessment of how technology can support LEAs in their daily efforts to address wildlife cybercrime. Section 1 includes a series of key definitions, as necessary background to understand the potential of the different types of solutions discussed. Sections 2 and 3 provide a non-exhaustive overview of a series of advanced tools that could help law enforcement officials to counter wildlife cybercrime.

Findings

In recent years, computer science has moved forward rapidly and mastered algorithmic models that are able, via data mining and machine learning, of making understandable sets of data that would otherwise be too vast or too varied to be comprehended. Thanks to these tools, today it is possible not only to identify wildlife sold on the internet but also to assess illegality. Moreover, tools derived from data science have the ability to shed light on existing links and tendencies online, and also to anticipate and predict future trends. These types of tools could yield greater information about the dynamics of illicit markets, the nature of the networks trading online, and ways to understand consumer demand.

Despite benefits of technology, human intervention still required to prevent wildlife crime on the internet

However, this report shows that it would be an illusion to think that technology alone, despite its appealing possible benefits, would be the ultimate means to counter wildlife cybercrime. At present, there is no one-fit-for-all, scalable, reliable, systematic and repeatable way to detect wildlife cybercrime automatically. Even though some tools reach a high level of accuracy when distinguishing between legal and illegal goods, their application is still restricted to a limited number of platforms (e.g. eBay) and products (e.g. ivory). Assessing automatically the illegality of an online advertisement selling wildlife still presents a major challenge due to primarily: the complexity of the wildlife trade regulations, covering many species and with several trade exemptions; and the lack of data available to “train” computers on a larger scale. Human expertise therefore remains key in this field. 

Automatic systems still generate a lot of irrelevant results, which demand substantial, often overwhelming, cleaning efforts from law enforcement officers. While technology appears potentially capable of simplifying and accelerating monitoring tasks, it appears that today it is rather changing the nature of these tasks. The large amount of time usually needed for manually monitoring the web is now spent cleaning up the results collected through automated monitoring and scraping. 

Furthermore, a critical challenge relates to the fact that both data mining and machine learning rely on the analysis of critical datasets. The amount of data needed to enable computers to locate and identify wildlife cybercrime is enormous and difficult to compile. Concerned entities amongst the LEAs and NGOs are unlikely to have the volume of data required, nor the technical means of collecting them. Specialized companies that have the means to acquire and process large amounts of data would appear to be the best option.

Preventing wildlife cybercrime will require coordination

Ultimately, using the highest potential of automated solutions to combat wildlife cybercrime will require a large multi-stakeholder collaboration and coordination involving all relevant stakeholders, such as police, customs services (environment and cyber units), academia, NGOs, and the giant tech-companies.