Mastercard experiences 200 fraud attempts per minute, and between 2018 and 2020 has declined USD14 billion in fraudulent authorization attempts. The stakes in these fraud cases can also be enormous, with individual incidents potentially costing millions in lost funds across multiple unprotected accounts.
It is not surprising, then, that Mastercard is leveraging advanced technological tools such as artificial intelligence (AI) to fight against fraud and avoid millions in potential fraud losses.
Training algorithms to detect fraud
Since 2018, Mastercard’s AI-powered Safety Net guards against large-scale fraud events, detecting and intercepting suspicious activities across the Mastercard network in real time. Safety Net monitors global authorization data for abnormal activity, taking account of spending patterns across channels, cards, and Bank Identification Numbers (BINs).
Mastercard’s AI team trains the Safety Net Algorithm by exposing it to transaction and authorization data, teaching it to autonomously identify potentially fraudulent transactions. Supervised learning techniques allow Safety Net to develop predictive models that determine if observed patterns and activities (authorization data, tracking location, velocity indicators, and profiling across transaction channels) are abnormal.
Making credit card transactions safer
If Safety Net determines that a transaction may be fraudulent, it can automatically decline the transaction, directing customers to self-serve tools to review transactions declined by Safety Net and to white list cards if card activity is genuine.
Should threats escalate to involve tranches or groups of potentially fraudulent transactions, Safety Net can temporarily block transactions across an entire channel. In cases of even further escalation across multiple cards or BINs, Safety Net automatically refers the issue to a Mastercard Tactical Response Team for emergency investigation.
Mastercard’s Safety Net has been put to the test in various situations and has helped issuers avoid losses in different markets. In one instance, it detected a massive merchant spoofing attack, and blocked USD26 million in fraud losses for issuers around the world.
Safety Net can detect ATM cash-out attacks, blocking some USD500,000 in potential fraud losses for issuers in Europe alone. it is also able to account for attacks depending on weaknesses in validation systems, such as exploits that target a gap in chip validation protocols.
Safety Net’s detection algorithm is even able to account for human error; upon detecting an issuer’s accidental deactivation of a chip cryptogram validation protocol, it blocked over USD11,000 in potential fraud losses.
According to Ed McLaughlin, Mastercard’s President of Operations and Technology, “Thanks to advances in artificial intelligence and machine learning, our technology can detect and address fraud more accurately and more quickly than before.”
Enabling transactions for a secure digital economy
Safety Net demonstrates the benefits of a holistic and dedicated approach to AI – especially when it is used not only as a productivity tool but to enhance an organization’s core competencies.
Indeed, Safety Net not only safeguards the information users share with each of their transactions, it helps them trust that the information they share is safe, no matter which unprecedented crisis is disrupting the processes and activities that keep entire economies afloat. It also shows that the human element – what ICT experts see as any system’s main vulnerability – can be removed from at least part of the equation without losing in reliability.
As Ed McLaughlin notes, “What is most important is to take the human out of the loop, so that they do not have access to personal data at the point-of-sale. It is about how powerful tools are used around AI to increase cybersecurity and protect consumers.”
In this sense, Safety Net lays the groundwork for safer cross-border and cross-platform transactions and embodies the potential of AI as a means of empowering consumers to address the challenges of an increasingly hyper-connected world.