How AI Is Transforming Credit Scoring and Lending

The Algorithm Will See You Now

When Mashreq Bank announced in February 2026 that its AI-driven credit engine had reduced loan default rates by 34% across its UAE retail portfolio, the reaction from traditional lenders was not surprise — it was anxiety. The Dubai-based institution now processes 82% of its personal loan applications without human intervention. It has become a case study in what happens when machine learning replaces legacy scoring models. Across the Gulf and emerging markets, a fundamental restructuring of creditworthiness is underway, and it is reshaping who gets capital, how fast, and at what price.

Beyond FICO: The New Architecture of Creditworthiness

Traditional credit scoring — anchored in repayment history and debt-to-income ratios — was designed for economies with deep formal banking penetration. In the GCC states, expatriate populations often lack local credit histories. Across emerging markets in Africa and Southeast Asia, an estimated 1.4 billion adults remain outside formal financial systems. These models have always been insufficient. Artificial intelligence is filling that void with an entirely different data architecture.

Saudi Arabia's Tamara, the buy-now-pay-later platform valued at $1.2 billion following its 2025 Series D round, now evaluates creditworthiness using over 3,000 data points per applicant — including mobile phone usage patterns, utility payment regularity, and e-commerce behaviour. The company disclosed in its Q1 2026 earnings call that its AI models approve 40% more applicants than traditional scorecards while maintaining a non-performing loan ratio below 1.8%. That sits well under the Kingdom's banking sector average of 2.3%.

In Nigeria, Moniepoint's AI underwriting system processed $4.7 billion in SME lending during 2025, using transaction data from its payment terminals to build real-time credit profiles for small merchants. The company's machine learning models assess cash flow velocity, seasonal revenue patterns, and supplier payment behaviour — variables that never appear on a conventional credit report. CEO Tosin Eniolorunda has called it "credit scoring for economies that never had the luxury of waiting 50 years to build bureau infrastructure." He has a point.

Gulf Sovereign Wealth and the AI Lending Stack

The capital flowing into AI-driven lending infrastructure from Gulf sovereign wealth funds and family offices has been extraordinary. Abu Dhabi's ADQ invested $180 million in India's Zest Money in late 2025, explicitly citing the company's proprietary language-processing models that assess loan applications submitted in 12 Indian languages. Mubadala's fintech portfolio now allocates roughly $600 million to companies building AI credit infrastructure across South and Southeast Asia. That is a significant concentration bet.

Kuwait's Markaz Financial Centre, the family-office-backed investment firm, launched a dedicated $250 million fund in January 2026 targeting AI lending platforms in frontier markets. Managing Director Fahad Al-Rajaan told investors that the thesis is straightforward: "Every market that skipped landlines for mobile phones will skip traditional credit bureaus for AI scoring. The arbitrage opportunity is in funding that transition."

Private wealth advisors in the region are increasingly directing clients toward these opportunities. Geneva-based Lombard Odier, which manages over $15 billion in Gulf family office assets, added AI lending platforms to its alternative credit allocation models in Q4 2025, noting in its annual outlook that the sector offers "uncorrelated yield in a rate environment that punishes conventional fixed income."

Regulatory Catch-Up and the Explainability Problem

The technology is moving faster than the rules designed to govern it. The UAE's Central Bank issued its Artificial Intelligence in Financial Services guidelines in March 2026, requiring all AI-driven lending decisions to be explainable — meaning institutions must be able to articulate, in plain language, why an applicant was approved or rejected. This "right to explanation" mirrors provisions in the European Union's AI Act and represents one of the most consequential regulatory developments in Gulf fintech this decade.

Compliance is proving expensive. Emirates NBD disclosed that adapting its AI lending models to meet the new explainability requirements cost approximately AED 120 million ($32.7 million) in the first quarter alone. Smaller fintechs face proportionally greater burdens. Sarwa, the Dubai-based digital wealth platform that expanded into AI-assisted personal lending in 2025, has hired 45 additional compliance engineers — nearly doubling its technical workforce — to build audit trails for its algorithmic decisions.

The tension is real. Black-box deep learning models tend to be more accurate but less interpretable than simpler decision trees. Bahrain's central bank, which has positioned the Kingdom as a fintech regulatory sandbox, is experimenting with a tiered approach: allowing more complex models for lower-risk lending categories while requiring full transparency for mortgage and business loan decisions above $500,000. It is an elegant compromise — if it holds up under pressure.

The Emerging Market Credit Dividend

The most consequential shift may be macroeconomic. McKinsey's April 2026 report on AI-enabled financial inclusion estimated that machine learning credit models could unlock $2.1 trillion in new lending across emerging markets by 2030, with the greatest impact in Sub-Saharan Africa, the Middle East, and South Asia. The consultancy projected that GDP growth in these regions could accelerate by 0.4 to 0.7 percentage points annually as previously excluded populations and businesses gain access to formal credit.

Egypt offers a compelling illustration. Fawry, the country's largest digital payments platform, launched an AI credit product in November 2025 that uses transaction data from its network of 350,000 merchants to offer working capital loans of up to EGP 500,000. Within four months, the platform had disbursed EGP 8.2 billion ($164 million) with a repayment rate of 96.1%. In a country where only 33% of adults have a bank account, those numbers demand attention. Few outside the region have noticed.

Jordan's Capital Bank Group, majority-owned by the Saif family's Social Security Investment Fund, reported in its 2025 annual results that AI-scored loans grew to 28% of its total book, up from 7% in 2023. The bank's risk-weighted return on these assets runs approximately 180 basis points above its conventionally scored portfolio. That is not a rounding error.

What Comes Next

The direction is unmistakable: credit decisions are migrating from human judgment and static models toward dynamic, continuously learning systems. For Gulf investors and family offices, the opportunity set extends beyond equity stakes in lending platforms to structured credit products backed by AI-scored loan pools — a market that Fitch Ratings estimates will reach $45 billion in issuance by 2028.

The risks are equally clear. Model bias, data privacy violations, and regulatory fragmentation across jurisdictions could slow adoption or trigger consumer backlash. But the economics are hard to argue with. When an algorithm can assess creditworthiness in seconds using data that didn't exist a decade ago, the question is no longer whether AI will transform lending. It is who will own the infrastructure when it does.