How responsibly deploying AI credit scoring models can progress financial inclusion

Egypt has made significant strides in expanding access to financial services over the past decade.

According to the Central Bank of Egypt, 74.8% of Egyptians held a transactional bank account as of December 2024, marking a 204% increase from 2016. The numbers are even more striking for women: 68.8% now have an account, up 295% in the same period.

This progress reflects a national commitment to financial inclusion, including regulatory changes such as allowing account openings with just a national ID. But it also reflects the contribution of innovative private-sector companies that are leveraging technology to unlock economic opportunity for millions.

Still, a bank account is only the first step. True financial inclusion means the ability to access, use and benefit from a wide range of services – from savings programmes to payments, credit and investment products. While access to these services is expanding, extending credit to underbanked individuals remains a key challenge facing financial institutions and governments seeking to broaden financial access.

Credit is a catalyst for long-term prosperity, directly facilitating wealth building by enabling individuals to invest in long-term economic opportunities.

Rather than paying rent each month to a landlord, individuals can take out a mortgage and make payments towards the ownership of their home. Families can finance their children’s education, and women can borrow capital, start their own businesses and gain greater economic freedom.

Yet for millions in Egypt and beyond, access to credit remains out of reach. Traditional credit scoring models rely on formal financial histories and manual underwriting is a resource-intensive and inconsistent gateway.

As a result, many banks and lenders are limited in their ability to extend loans to new-to-credit and underbanked populations. This chicken-or-the-egg dilemma most heavily impacts low-income, rural and financially excluded communities, including women and youth.

Innovators are increasingly turning to artificial intelligence (AI) to tackle the limitations of traditional financial services models.

In economies where formal credit histories are scarce and credit bureau coverage is limited, AI-based models offer a compelling alternative. What sets these models apart is their ability to interpret not just individual activity, but also social and behavioural networks.

By integrating diverse, real-time data points, AI-powered credit scoring models can improve approval rates, reduce risk and extend credit to those previously overlooked by conventional systems.

This is a promising development for financial inclusion – especially in contexts where traditional methods simply do not scale.

MNT-Halan, Egypt’s fastest-growing fintech company and MENA's first fintech unicorn, has developed an AI-powered alternative credit scoring engine that uses behavioural and transactional data to evaluate users – many of whom have never interacted with formal credit systems before. Through this proprietary model, MNT-Halan has automated more than 50% of its loan approvals and achieved a 60% approval rate for previously unscoreable users.

The solution leverages data from the Halan superapp which provides payments, savings, investment and e-commerce services on a single digital platform. This integrated environment generates a wealth of user data – purchases, payment history, interactions with the app – that feeds directly into the AI engine.

By observing how users interact across these services – how frequently they make purchases, repay balances or engage with in-app tools – the model builds dynamic and individualized profiles.

This multi-dimensional model enables MNT-Halan to assess risk far more accurately than traditional systems, even in the absence of formal financial data.

Take, for instance, a previously unbanked individual who began using the Halan app’s e-commerce feature to buy wholesale groceries. Based on this behavioural data, the AI scoring engine was able to assign a credit score and approve the user for a consumer finance limit.

From there, the individual was approved for a Halan Card and later began to use its secured investment products. This was a full financial journey that is enabled by AI and driven by data.

Implementing AI in high-impact areas like credit scoring requires robust risk management frameworks. If not designed carefully, these systems can reinforce the exact inequalities they intend to alleviate.

By proactively developing responsible AI principles and conducting regular audits, companies can ensure that AI is deployed fairly and responsibly. MNT-Halan utilizes five principles in its employment of AI:

Crucially, users opt in to data sharing – and can opt out at any time. This opt-in model prioritizes consent and trust with the products’ users.

The implementation of AI credit scoring models is already helping millions in Egypt access financial services they’ve long been excluded from. But the opportunity doesn’t stop at national borders.

According to the UN Economic and Social Commission for Western Asia, 64% of adults in the Arab region remain unbanked – and only 29% of women have an account. Since establishing operations in Turkey, Pakistan, and the United Arab Emirates last year, MNT-Halan is working to expand access to credit across the Middle East and North Africa region.

Achieving true financial inclusion for these individuals will require more than simply providing bank access. The intelligent implementation of AI can help ensure that inequalities in data availability do not reinforce existing disparities in financial services access, providing a crucial tool for promoting economic prosperity for those who have been traditionally excluded from formal economies.