The problem of fraud in e-commerce and cashless payments in the MENA region and worldwide – and how Artificial Intelligence can help.


An interview with Joshua Bower-Saul, CEO of Cybertonica

Joshau Bower-Saul
Joshau Bower-Saul

Cybertonica is a young company born through collaboration among FinTech specialists. It uses machine learning and specific data about transaction behaviours to build real-time dynamic responses to fraudsters.

Joshua Bower-Saul will be speaking about how AI and Machine Learning can support fraud management at PayExpo MENA in Dubai at 1055 on Wednesday 7 December

  1. Can you please quantify the scale of the problem of fraud in e-commerce and cashless payments worldwide and in the Middle East? If possible, how does the Middle East compare globally in its approach to prevention of cyberfraud?

The Middle East and several countries in the region have a deep consciousness of the problem and are subject to limits on their online trade around the world that exist due to the risks of cybercrime and fraud.

It is known to us at Cybertonica that limitations exist in Turkey and other countries on selling into or buying from cross-border countries such as Iraq and Iran due to the volatility of the security environment and the prevalence of Card Not Present fraud. Turkey and India (not the Middle East directly, but similar markets, have called for secondary authentication on 100% of transactions.  

So as transactions go down and basket loss goes up, while fraud remains low, companies are losing business. In some cases, transactions are simply “quarantined” by IP or regional origin of the transaction with no other basis but the “expectation” that fraud will be higher there than elsewhere.

We heard from one merchant that they stopped all transactions between Israel and Russia above $300, simply because they could not model the fraud or have access to card data that would reassure them on their liability exposure. This represented 20% of their business. With Cybertonica these problems can be reduced or eliminated due to the manner in which we manage data across borders and the analytics we can apply to individual verticals or user-device-channel combinations automatically. We can also remove controls extensively in secondary authentication.

  1. How do AI and Machine Learning help to increase security in this issue?

AI is not a magic word. You have the same problem of smart systems and dumb systems that co-exist under this label. We feel our ML and AI are superior because they have been tested in tough markets like Russia and have survived and benefitted customers with 10s of millions in additional revenues.  How we do that is by using the AI to beat those using AI to “deconstruct” the rule sets on the banking and payments channels. Yes, sadly AI can be used by the bad guys too. What we are building is a platform that brings access to AI defences to the enterprise in all phases of banking and digital purchasing.

All channels are also covered and in our roadmap for next year we have full mobile security as well (remember that maybe as many as 25% of Android phones are affected by hijacking at one time or another). What AI and ML do is make the management of massive flows of data across all these node points and devices and interfaces automatic and monitoring and analytics visible (using for example our proprietary data and system visualisation engine) in a way that focus on the business impact of risk-management decisions.

In the old rule-based systems, a cumbersome interface between databases and individual instances of anomaly makes for a troublesome and costly installation, reporting and delayed fraud response. Our system in real time can handle billions of events per day on its AWS cloud cluster and provides minute-minute monitoring so patterns of fraud and systemic abuse by criminals are immediately detected and signalled out to the merchant, payment platform or other authorities if legal mandate exists. On 100 million transactions for a client we lowered the chargeback rate by 70% and still increased sales by 30%.

  1. What are the key lessons that delegates to PayExpo MENA will take away after listening to your presentation about how AI and Machine Learning can support fraud management?
  • The rules based systems of the last 10 years are now being overrun by criminals using social engineering and AI/ML to “game the algorithms (see Tesco Bank). You need a system that can manage billions of events automatically and communicate in a meaningful way to business owners and to fraud managers and merchants.
  • That as channels for payment broaden and deepen and CNP becomes a prevalent form of payment by 2020 in this and other regions, a hub like Cybertonica will be a gateway to fraud-free trading.
  • That the danger to banking and payment systems also comes from WITHIN and that is why an external hub that is “taking the pulse” while protecting personal data in a compliant way is a useful ally.
  • That the use of Risk Based Authentication (RBA) in the future will be the standard, no matter what is planned for the next phase of secondary authentication, since only a RBA approach can manage the huge volumes without causing system and customer pain.
  1. What are the next steps and developments that you expect in the use of AI and cybertech to prevent online fraud?

Substitution of most rules based integrated systems by SAAS and onsite quasi SAAS offers like ours.

  1. Apart from the increased adoption of AI and machine learning, what should governments, MNOs, banks and IT managers being doing to further help to prevent cyberfraud and theft?

The primary thing they need to do is challenge their assumptions about how fraud works and about the fact that this is now in itself a BIG business, using technical resources against the legitimate world of eCommerce and digital banking to take huge payouts illegally from under-prepared systems. They should not panic, but they should move quickly to go to the cloud for help (because in fact the Cloud is MORE secure when using AI than most banking systems today). 

There should be more education about the reality of the situation and more regulatory support for new systems such as RBA to work alongside mandated systems to show their innovation and its capability.

Joshua Bower-Saul will be speaking about how AI and Machine Learning can support fraud management at PayExpo MENA at 1055 on Wednesday 7 December

PayExpo MENA:

A wide range of new features and conference streams shaping the future of payments will be unveiled at this year’s PayExpo MENA, taking place at The Intercontinental Festival City, Dubai on 5-7 December.

Complimentary passes are available for those involved with specifying payments solutions at:  Retail Banks; Investors and Venture Capitalists, Merchants; Mobile Network Operators, Gaming Executives, Transport Operators, Hotel and Leisure Executives, Start-ups and Non-Government Authorities.

Registration is now open at:

Etisalat is the MNO Partner, The Middle East Council of Shopping Centres is the Retail Partner, Marka VIP is the Online Retail Partner, Bank Dhofar is the Host Bank, The Cribb is the Start-up Incubator Partner and TURN8 is the Venture Capitalist Partner.

About Clarion Events:

Founded in 1947, Clarion Events is one of the oldest independent event organisers in the UK. More recently the firm has developed an international portfolio of brands and now has interests in a multiple industry verticals and delivers more than 200 events each year globally.

About the Payments Portfolio:

Clarion organises a global series of conferences and exhibitions in the Cards and Payments industry to enable banks, retailers, transport operators and payment processors to keep up with the rapidly changing trends in consumer spending behaviours and payment solutions. The Portfolio consists of PayExpo and The Transport Ticketing World Series.