Skip to content
Artificial intelligence (AI) is still a buzzword for many in anti-fraud circles, but its application and refinement are accelerating. Brandon Daniels, President of Global Technology Markets at Exiger spoke to Fraud Intelligence about how anti-fraud professionals can realise the potential of AI to uncover fraud.
Brandon Daniels said companies and governments needed to learn how criminals deploy Al to carry out fraudulent activities, and companies wanting to fight fraud with Al needed to follow tactics inspired by fraudsters. “Criminals have surpassed simple statistical algorithms” he observed, “They are using deep neural networks of data collection and analytics to assess your traffic, identify your likely credit card types, and automatically generate sophisticated phishing ‘bots’ to prompt you to expose your personal details. These data attributes are then reverse-engineered to execute latent fraudulent transactions on your behalf.”
Companies fighting fraud with Al need to take note: “Analysing where there is a spike in similar buying activity across multiple disassociated customers’ transactions takes fast and large-scale data analysis,” best executed by Al, which can help spot latent risk. Alternatively, by deploying those same neural networks used by fraudsters a company can identify voice prompts that are easily understood by a human but are “consistently failed by a [non-AI] machine,” he said. Fraud detection and suppression is most powerful when you use machine learning training to determine the data points that are difficult for a system to analyse, because they consistently sit outside those usually identified in assessments by non-AI systems. “You have to fight fire with fire,” Daniels concluded.
