'How machine learning will transform the financial crime and fraud detection equation', Michelle Weatherhead, General Manager of Commercial Solutions ANZ, BAE Systems
"Machine learning represents a new era in the fight against financial crime for Australian businesses, with the technology backed by the cloud, it is now more accessible and affordable for at risk companies than ever before.."
The cost of financial crime is significant for Australian businesses – from both the cost of compliance and the huge fines associated with non-compliance. Machine learning represents a new era in the fight against financial crime for Australian businesses, with the technology backed by the cloud, it is now more accessible and affordable for at risk companies than ever before.
The use of this form of artificial intelligence (AI), computer systems which are programed to learn to identify patterns in data and make educated predictions based on that analysis, is increasingly being used by businesses around the world.
Globally, the customers we work with often detect twice as much fraud using a combination of complex analytical and machine-learning techniques. So what makes machine learning more than just another buzzword from the technology industry?
From the big banks, to smaller financial institutions to what is commonly referred to as ‘gateways’ – real estate agents, lawyers and currency exchange institutions –all are under threat from criminals attempting to defraud them and their customers, or move illicit money in ways and to places that put the public at risk.
Laundered money represents between two and five per cent of global Gross Domestic Product (GDP), according to the UN. The Australian Institute of Criminology has estimated the cost of fraud to the economy is $8.5 billion annually, with criminals increasingly using sophisticated methods to dupe legitimate businesses both in person and online. For businesses of all industries and sizes to avoid the financial and reputational risks associated with financial crime, a combination of skilled staff, the right processes and technology in place is crucial.
Skilled technology and compliance staff are increasingly difficult commodities to come by. The Australian Computer Society and Deloitte has projected a gap of more than 100,000 technology workers in the next five years, and declining rates of ICT graduates from universities.
With a serious skills issue, machine learning presents an opportunity to support the staff in place and deliver capabilities beyond what staff can currently provide. As financial services technology increasingly speeds up the recognition and settlement of transactions, the detection and assessment technology will only keep pace through machine learning.
Machine learning works in the same way as complex analytics performed currently. Each transaction is given a fraud score which can help determine the chance of the transaction being illegitimate. The program’s model is customised to the specific data and then updated periodically to identify new fraud patterns. This exceptionally fast and dynamic system of detection helps reduce the number of false red flags and support staff in correctly identifying more undesirable activity, more accurately.
That’s not to say the machines will make jobs in the sector redundant. Machine learning does not make their roles in identifying suspicious activity pointless, it simply does the heavy lifting for them, faster, freeing people up for tasks that add higher value. The technology can only work effectively once there is someone using the data in innovative ways to fight fraud.
Beyond the significant skills gap, the cost of compliance is rising rapidly for financial services institutions. From 2007-2013, CEB found the world’s six largest banks saw their compliance costs more than double from $34.7 billion to $70.1 billion.
Machine learning is critical in bringing the cost of compliance under control and reducing the financial and reputational impact of fraud. Concur’s research recently found card issuers and banks alone can save up to $12 billion annually via machine learning technology.
By accessing machine learning and making it an integral part of financial crime detection processes, at risk Australian businesses will be accessing technology and expertise which augments what they currently have in place to make it more effective than ever before.
While we may never be able to fully stop financial crime, we can certainly use latest and best technology to catch more criminals and reduce the burden on businesses and consumers.