Combating Wildlife Trafficking and Abuse in Social Media and E-Commerce

Social media and e-commerce have transformed the trade in endangered species over the last two decades as social media platforms and international law enforcement grapple with this fast-changing dark market. Technological innovation is at the forefront of tackling illegal online activity as well as more traditional illegal offline trading.

New technological solutions are being employed to identify animals and perpetrators and provide evidence in prosecution to analyse patterns in criminal activity that inform initiatives to reduce the impetus or ability for people to engage in crime against wildlife or captive animals.

CITES and the Growth of Digital Wildlife Crime

Google.com doesn't filter exotic animal for sale content of any kind.

In 1975 the Convention on International Trade in Endangered Species for Wild Flora and Fauna (CITES) came into force, regulating the trade in wild animals across borders. CITES was agreed upon by 80 national governments and established as a control mechanism to protect 5,000 endangered animal species and 28,000 plant species. Each country is responsible for enforcing rules under CITES, and many of these countries create additional legislation or mechanism to protect particular species of national interest.

Whilst the illegal wildlife trade continues in traditional non-digital markets, traders now use social media and e-commerce to buy and sell endangered species. In China, more than half of the trade in elephant ivory takes place on e-commerce platforms.

Wildlife trade on e-commerce sites in China, with a focus on mammoth ivory: A Rapid Assessment November 2021

Digitising Image Recognition and Context Analysis

Monitoring the illegal wildlife trade on social media involves various challenges, including processing the sheer volume of data and effectively providing cover for criminal operators. Data law enforcers investigate leads and monitor known people and organisations, but big data processing is still relatively recent and developing fast. As a result, traders often hide from search functions by using code words of greater or lesser sophistication to cover illegal transactions or simply labelling posts with generic terms such as "for sale".

Data processing can be automated so that data is mined directly from social media platforms, filtered and cleaned so that only data with higher relevance is presented for further investigation. Neural networks, also known as deep learning, are a machine learning process using a layered structure of interconnected nodes inspired by the design of the human brain. They employ a set of algorithms that can be trained to recognise and classify species and wildlife products — like a pangolin or rhino horn — from images contained in social media posts. They do this while also processing visual context, for example, that the pangolin is in a cage in a marketplace or that other species or people are present. In addition, neural networks can be trained to look for additional information in videos, including audio data such as specific bird calls.

The visual and audio data in the photo or video can be evaluated alongside contextual data, including any associated written content and metadata such as where the image was created and published, by who, when, and using what equipment. Some aspects of machine capability are far beyond any human — for example, the ability to operate in 10+ languages simultaneously.

A Powerful Tool for Investigators

Longer-term investigations aim to crack codes unobtrusively, thereby accessing criminal networks to target keystone personnel. Neural networks are a potent tool for wildlife crime investigators for immediate interceptions and long-term covert operations.

Enrico Di Minin and Christoph Fink wrote a paper proposing a system that uses machine learning to investigate illegal wildlife trade on social media platforms using automatic content identification methods. For example, a neural network was trained to identify Twitter posts with the word "rhino" in any of 19 languages containing images of rhinoceros species.

The automated filtering reduced the number of images a human expert would have to review by over 90%. In another study, researchers detected illegal elephant ivory items for sale on an e-commerce platform using metadata.

Each social media platform has its process and philosophy for removing posts, and its relationship with wildlife crime investigators. As a prominent example, 2.9 billion people use Facebook monthly — a company that joined the Coalition to End Wildlife Trafficking Online in 2018. The aim at the time was to reduce the illegal wildlife trade online by 80% within two years. However, Facebook is still a thriving marketplace for illicitly selling and purchasing endangered species.

Facial Recognition and Real-Time Tomography

ChimpFace was one of the finalists in the Con X Tech Prize in November 2018. It will become a facial recognition application capable of identifying individual wild chimpanzees. Facial recognition is an important tool for wildlife management and identification of animals showing up in illegal content online. Real-time tomography — a form of 3D scanning — can be used to identify animal products such as ivory, even when disguised or altered.

Looking Ahead

Technological innovation is rapidly advancing our ability to detect, monitor, and prosecute wildlife crime online. But technology alone is not enough. Platform accountability, international cooperation, and sustained political will are equally essential to turning the tide on the digital wildlife trade.

Wild Around supports the development and deployment of conservation technologies that address these challenges — connecting innovators, investors, and enforcement communities in the fight to protect the world's most vulnerable species.