Australian Government Experimenting with AI to Stop Wildlife Smuggling

While numerous countries worldwide have to cope with wildlife trafficking, none feel the pain like Australia, but the introduction of new AI technologies could soon change the game. What challenges does wildlife trafficking introduce, what does the new AI technology allow for, and could AI be used in other border control applications?

What challenges does wildlife trafficking introduce?

As much as we all find border control and customs a nuisance, it exists for good reasons. In the case of people, only those who want to enter on legitimate terms must be allowed to enter, and in terms of goods, it is important that goods are correctly taxed and authorised to be sold in the country of entry. At the same time, customs and luggage checks are essential for seizing illicit content and materials, including drugs and weapons. 

But there exists one type of cargo that can prove to be exceptionally difficult to identify and seize; Wildlife. As strange as it may seem, numerous countries around the world enforce strict controls on what animals can pass through borders, and there are various reasons why this control is essential. 

In the case of Australia, preventing wildlife from entering the country is essential as Australia is an island, and this means that its ecosystem has developed independently from the vast majority of the world. Therefore, introducing non-native species can cause significant pest issues and environmental damage. An excellent example of this is how the Red Squirrel can only be found on the Isle of Wight (in the UK), as grey squirrels eat acorns at an earlier time of the year (therefore starving red squirrels of food). 

Wildlife controls are also needed for animals leaving the country as most animals that are smuggled out of countries are endangered. The high value of exotic animals combined with their endangered status can see smugglers destroy native populations, and trying to find these animals in transit can be difficult. Not to mention that smugglers are by their nature criminal, and the money raised by smugglers undoubtedly contributes to more crime and untaxed income (we all know that untaxed income is far more likely to be investigated). 

Australian researchers experiment with AI and X-Rays

Current, the method for detecting suspicious cargo is to utilise x-ray imaging to see inside packages and containers. Despite the ability to show the internals of a closed box, x-ray imaging still requires human eyes to make sense of the data. Furthermore, the vast amount of cargo that goes through country borders is so mind-boggling officers only have moments to observe the inside of a container and then make a decision.

Recognising the challenges faced by the Australian Government, researchers in Australia have recently demonstrated a new imaging technology that leverages AI to automatically identify the contents of scanned items. In order for the AI to determine the presence of illicit content, the package under investigation is first scanned in 3D using an airport customs X-Ray CT scanner. Once 3D models of the scanned item have been generated, they are fed into an AI that has been trained against a reference library of 3D objects (in this case, the researchers trained their AI to identify 13 species of lizards with over 296 scans in total). 

According to the researchers, their AI identified animals with an 82% success rate and a 1.6% false-positive rate. While some may see the 82% success rate as low, it represents a significant capability as the entire system is automatic. The packages that are not identified would likely pass a manual scanner regardless, and such AI systems can be improved over time. In fact, the very act of checking packages to confirm the results of the AI would only reinforce its ability to identify illicit materials in the future.

How could AI technologies help with border control in general?

It is truly amazing to see how far AI has come in the past decade. What started out as research papers demonstrating basic pattern recognition have become full-fledged intelligent systems capable of extreme pattern recognition in datasets that otherwise seem entirely unrelated. Furthermore, the use of AI in law enforcement could provide numerous advantages to authorities, including high-speed identification of individuals, tracking stolen goods, detecting unusual behaviour, and possibly even anticipating crimes before they occur.

In the case of border controls, AI could be extended well beyond wildlife smuggling; just about any item could be flagged by such an AI, including weapons and drugs. The use of such technologies may even be able to identify sensitive components, such as high-end semiconductors being shipped to foreign nations that have otherwise been banned from receiving such parts. 

Overall, AI will continue to play an increasingly more prominent role in our lives, but while it can provide massive benefits, it can also come with challenges. Thus, when using AI in law enforcement, we must ensure that it is being used to aid investigations while not treating every citizen as a suspect.