Article

How the video gaming industry is using machine learning to fight fraud

Le Raine Hendrik
Le Raine Hendrik

Content Writer

alt text for image

Artificial intelligence (AI) has long been integrated into game design, and it’s now at the forefront of some of the industry’s most cutting-edge techniques. Machine learning—an area of AI where the computer is able to learn and adapt to new data without being explicitly programmed—now offers game developers a massive advantage in one significant area: reducing payment fraud.

An uphill battle

Given the fact that it’s worth a hundred billion dollars[1], the gaming community is a lucrative target for cybercriminals. Online gaming platform Steam claimed in 2015 that around 77,000 of user accounts get hacked and stolen every month. “Enough money now moves around the system that stealing virtual Steam goods has become a real business for skilled hackers,” the company said[2].

Adding to the appeal for fraudsters is the fact that the gaming industry is particularly vulnerable. For one, the industry is not as heavily regulated as other sectors such as finance and healthcare, and game makers don’t need to comply with a standard set of cybersecurity protocols.

Cybersecurity also simply isn’t given the priority or attention it deserves in the industry. Between the pressures of building in additional features, ensuring that the code is bug-free, and getting the games to market as quickly as possible, creating a secure infrastructure is often overlooked. Unless there’s a large-scale hack or malware attack, game developers generally regard billing fraud as the cost of doing business.

Even with security measures in place, however, game developers still run into challenges. Traditional fraud solutions operate on a set of rigid rules, which run the risk of false positives – that is, flagging genuine customers as fraudulent ones and disrupting their gaming experience. Since a large amount of goods in the gaming industry are digitally delivered, it’s impossible for game publishers to manually review and approve these suspicious transactions. The outcome is either a frustrated customer, who wants to pay for the product but gets locked out, or a fraudulent transaction slipping through.

A game-changer for fraud management

This is where machine learning can make a huge difference. Fraud solutions based on machine learning run autonomously, freeing up time for game publishers while facilitating a seamless transaction. They also offer a limitless degree of flexibility, growing more accurate and effective over time.

The machine learning model receives a new data set in real-time whenever a gamer takes action on your app or website, which it compares to the previous data set and evaluates. A new risk score is then recalculated, taking into account the player’s behavior over the entirety of the machine learning’s history of it.

A machine learning fraud solution, such as the one from Payvision, offers the complete package. It evaluates thousands of signals, such as whether the player’s historical and current purchasing and browsing behavior make sense, what their payment methods are, whether their location details match up, and more. The program can then make more intelligent predictions about which customers can be trusted or not, but more importantly, it’s able to continuously learn and fine-tune its assessment in real-time for maximum accuracy, and stop fraudulent transactions before they happen while improving conversion rates for game publishers.

Artificial intelligence is breaking new ground in the gaming industry, simplifying tedious design work while surpassing human creativity. More importantly, it’s creating a safer game environment for gamers and developers alike to enjoy.

[1] (Newzoo)

[2] (Steam)