Trust the Machines – Let Them Help You Fight Fraud
Anyone who’s seen The Terminator, The Matrix or more recently Ex Machina could be forgiven for thinking that the machines will inevitably turn on us. The reality away from the big screens is that artificial intelligence and machine learning is all around us and is benefiting our lives in more ways than we had perhaps appreciated.
For example, Spotify is using machine learning to evaluate user preferences and suggest increasingly accurate new music recommendations, while car manufacturers are using it to create the first road-safe driverless cars. Already we’re depending on the power of machine learning for important personal issues too; medically it’s being used to research disabilities and prevent avoidable hospitalisations, while in an increasingly threatening global fraud landscape, it helps protect us every day.
Fraud threats are evolving constantly as criminals devise more inventive ways of bringing down online businesses. Last year, Risk Ident conducted a survey which found that online merchants had a rather long cycle when it came to adjusting their fraud prevention rules: 42% change them once a quarter, 26% once every six months and 22% only readjust them once every year. This is dangerous, as fraudsters don’t sit still or update their tactics every quarter or year; they evolve their methods constantly, probing for weaknesses to exploit.
Fraud prevention should be able to intelligently evolve to meet these changeable threats as they arise. In the near future we can expect to face the following challenges:
In 2016, Yahoo fell victim to one of the most devastating breaches we’ve ever seen, reportedly leaking data from more than 1 billion user accounts, while UK telecoms company TalkTalk was ned a record amount by the authorities for allegedly failing to apply “the most basic cyber security measures” for 150,000 customers. A recent Office for National Statistics (ONS) survey estimated that the UK saw 3.6 million cases of fraud and 2 million computer misuse offences last year, making fraud “now the most commonly experienced offence.”
This trend will only continue in 2017, where hacking will target private user information, including names, email addresses, telephone numbers, dates of birth, passwords and security question answers. Collectively we need to identify such threats early and minimise any potential damage.
Snooping on social media
Social media leaves our personal information vulnerable to fraudsters who use Google, Facebook, LinkedIn and Twitter as research facilities. The number of identity the victims rose by 57% between 2015-2016, according to non-profit organisation Cifas, who attributes this sharp rise to information made public on social sites. It has never been so easy to piece together information on a specific person, and sometimes no technological expertise is needed: fraudsters simply take advantage of people sharing too many details publicly, which should be kept private. Spreading awareness about the dangers of over-sharing will help cut the levels of damage fraudsters can cause.
Every internet user has a lot of different online accounts. Instead of using unique passwords for every service – as experts suggest – most of us take the easier route and re-use a few favourites. But the fraudsters are aware of this weakness and are using it to their advantage. For example: shopping on the black market in the dark corners of the web, fraudsters can buy usernames and passwords and use them to try multiple other accounts online. We expect fraudsters to attempt even more in 2017.
Booming bots attacks
Disruptive smart so ware can generate spam, vandalise information on Wikipedia or try to influence opinions on social media. But bots are helping fraudsters as well. For example, when it comes to ticketing, bots are often able to order tickets faster than real customers. The number of bots, and the level of their intelligence, will continue to increase in 2017, so expect the ticket black market, among others, to grow.
The number of Europeans regularly using a mobile device for payments tripled from 2015 to 2016 (18% to 54%), according to Visa. Through a mixture of contactless, online and in-app payments, more of us are using mobile devices to complete transactions. However, fraudsters are using these new channels to exploit weaknesses. Often they will use multiple portable devices, alongside other masking techniques, in attempts to avoid triggering fraud alerts.
Man and machine – a perfect team
Using data science and machine learning, online merchants can create intelligent algorithms capable of detecting connections between individual transactions as well as unidentified fraud scenarios. While fraudsters seek to conceal their locations, mask their identities and trade payment card details or other personal information online, machine learning technology is able to find patterns, calculate risks and halt their activities – in real-time.
Fraud managers are indispensable in this process. A human being with years of experience fighting fraud can never be replaced by a machine, but a combination of the two entities can produce fantastically accurate results. Domain experts know their fraud problems best but they need scalable so ware to help. By constantly feeding their knowledge on the context and causes of fraud into the machine, the system can evolve continually. Fraud managers can therefore scale their fraud protection system by teaching the machines to help monitor for illegal activity.
This new strategy is now being taken up by merchants across the world and will become ever more critical in helping turn the tide of online fraud. Once the machines have been intelligently taught, they can become a scalable, accurate and consistent weapon to help us terminate fraud threats before they unleash chaos across the online world. The machines are not turning on us – they can make our lives better and help merchants take on fraudsters.
Download the original article (PDF)