AI breaks the mould for cybersecurity in South Africa

Think Progress Team

Thursday 19 January 2017

Think artificial intelligence (AI) and you most likely picture self-aware machines running amok in sci-fi movies. In reality, the technology is coming on-stream in a far more understated – and helpful – manner.

Digital security in an increasingly connected South Africa is an issue only growing in complexity. However, while AI is a still-maturing technology, it is developing fast and is particularly suited to the demands of cybersecurity. As new insight highlights, those demands are multiplying in South Africa and across the southern half of the continent.

Ericsson’s Sub-Saharan Africa Mobility Report predicts smartphones penetration in the region will hit 80 per cent by 2022, with subscriptions set to rise by 21 per cent each year until then.

The same report, detailed on ITWeb, also notes that the Internet of Things (IoT) will play a larger role in the delivery of innovative business models in the region. Connected devices will increase from 11 million in 2016 to 75 million in 2022. When you consider that humble kitchen appliances can be a ready source of Wi-Fi passwords, the security risks are obvious.

Leaving the desktop behind

Businesses need to respond to these security challenges. Arthur Goldstuck is an internet and mobile technologies expert who recently described how the South African digital security landscape might look over the next few years.

Speaking at ESET Security Days, Goldstuck revealed that South Africa is moving from mobile first to mobile only, given the explosion in smartphone use and the relative lack of backup devices at home, such as laptops and tablets.

Goldstuck told the conference that over the next 10 years there are three certainties: “By 2020, your businesses will be mobile and therefore digital. Your customers will be mobile. And both you and your customers will be more vulnerable than ever before.

“The challenge will increasingly be to protect both yourself and your customers in this variable future, where your customers and users have different demands, experience and needs.”

AI is adaptable, scalable and sophisticated enough to play a big part in providing that protection. Goldstuck went on to say that he believes AI will offer a truly capable solution by 2021, when internet users in South Africa are expected to number almost 25 million. By that point, the technology should have advanced to the point where users will detect little difference between a chatbot and a human agent.

AI in action

One indicator of how AI could be integrated into cybersecurity strategy comes from the banking industry. MyBucks is a fintech firm that has established a foothold across Africa, operating in Uganda, Mozambique, Tanzania and Kenya. Its aim is to provide digital banking services to people who have been traditionally underserved by brick-and-mortar banks to “break the ongoing cycle of poverty,” said CEO Dave van Niekerk in a MyBucks press release.

Across a continent 54 countries strong, regional differences proliferate. That means the way money is sent, lent and spent can’t be managed by a one-size-fits-all solution. While AI can help with data gathering to mitigate that, van Niekerk believes it can play a much more significant role: predicting if someone can pay back a loan.

A proprietary credit-scoring and decision-making system analyses a customer’s financial behaviour and forecasts their probability of defaulting against future repayments. Transactional data and employment information feed into algorithms that, so far, have accurately made 1.5 million predictions.

The company is also looking to AI to fight fraud: “[We are] currently also investigating ways to use classification or grouping algorithms within AI to identify fraudulent behaviour and are in the process of developing facial recognition algorithms as part of our fraud prevention system,” van Niekerk said. “Here the development of our internal AI fraud detection module supplements the functionality provided by our credit decision and scoring system.”

Banking between the gaps

While MyBucks is on something of a pathfinding mission in applying AI to cybersecurity in Africa, others are on the same page. Tina Blazquez-Lopez is counsel at law firm Pillsbury, co-leading its Africa initiative. She told IDG Connect that AI is adaptable enough to combat very specific types of fraud.

“There are significant leakages in the inter-bank transfer system with numerous phantom transactions being made on a daily basis,” Blazquez-Lopez said. “There are also a number of ghost workers who live on the continent and receive payments through a flawed banking system.”

The South African tech sector has already made great strides in AI machine-learning techniques. Yet as it ponders the digital security challenges posed by a nation that accesses the internet largely on its smartphones, it could do worse than look beyond its borders. AI advances made in countries where digital banking has greater reach than traditional high-street branches are already showing the way forward.