How AI Is Transforming Credit Scoring and Lending

When Mal raised $230 million in seed funding in January 2026 — one of the largest early-stage fintech rounds globally that quarter — the Abu Dhabi-based platform sent a clear signal about where Islamic finance is headed. Mal is not a digital bank with a Sharia-compliant wrapper bolted on. It is an AI-native institution, built from the ground up to deploy machine learning across every layer of its credit and risk infrastructure. That distinction matters more than it might initially appear. Across the Gulf, Central Asia, and Southeast Asia, artificial intelligence is quietly rewriting the rules of who gets credit, at what price, and on what terms — with consequences that will reshape consumer lending and institutional capital allocation for the decade ahead.

Beyond the Credit Bureau: What AI Actually Changes

Traditional credit scoring is, at its core, a backward-looking system. It relies on formal financial history — loan repayments, credit card usage, bankruptcy filings — to predict future behaviour. That model works reasonably well in markets with decades of bureau data and high banking penetration. Almost everywhere else, it fails. In Indonesia, Nigeria, and Pakistan, a significant share of the adult population carries no formal credit history whatsoever, despite earning consistent incomes, running viable businesses, and managing household finances responsibly for years. The bureau simply never saw them.

AI-based credit models break from that constraint by drawing on alternative data: mobile phone usage patterns, utility payment behaviour, transaction velocity, geolocation consistency, even social graph analysis in some emerging-market deployments. The result is a credit signal generated in minutes for individuals who would have been invisible to any conventional bureau. Mal's expansion discussions with regulators in Bangladesh, Indonesia, and Pakistan are not coincidental. These are precisely the markets where alternative data-driven lending creates the most immediate addressable opportunity.

The numbers tell a complicated story — but one that ultimately favours the bulls. Lenders using AI-enhanced underwriting models in Southeast Asia have reported default rate reductions of between 15 and 30 percent against traditional scorecards, while simultaneously approving a materially higher share of first-time borrowers. Broader inclusion with tighter risk management. That combination is the core commercial proposition pulling serious capital into the sector.

Open Banking as the Data Infrastructure Layer

AI credit models are only as powerful as the data feeding them. Which is why Saudi Arabia's central bank SAMA granting its first live open banking licences in March 2026 is not merely a regulatory milestone — it is the activation of an entirely new data infrastructure for the Kingdom's lending sector. For the first time, fintechs and banks can exchange real-time customer financial data through standardised APIs, with customer consent. Credit decisions that previously required days of manual document verification now happen in seconds, drawing on verified, live transactional data.

The implications extend well beyond consumer lending. Small and medium enterprises — long underserved by Gulf banking systems — can now present a complete, real-time financial picture to prospective lenders without paper submissions or weeks-long review cycles. For family-owned businesses across the GCC, which form the backbone of non-oil private sector activity, open banking combined with AI underwriting could meaningfully cut both the cost and the time required to access working capital. That is a significant shift, and most observers outside the region have yet to price it in.

Tabby's trajectory illustrates how rapidly this infrastructure gets commercialised once the regulatory plumbing is in place. Having reached a $4.5 billion valuation following its October 2025 secondary share sale and secured a Stored Value Facilities licence from the Central Bank of the UAE, Tabby is no longer a buy-now-pay-later company. It holds customer funds, issues payment cards, and is building money management tools — all of which generate the proprietary transaction data that feeds its AI-driven credit models. CEO Hosam Arab has been explicit that the company's ambitions extend well beyond instalment payments. The licensing and data accumulation strategy makes that trajectory entirely credible.

Sharia Compliance in the Age of Algorithmic Lending

For markets across the Gulf and the broader Muslim world, AI-driven lending carries a distinctive additional layer of complexity: compliance with Islamic finance principles. Conventional interest-based credit structures are impermissible under Sharia law, which requires profit-and-loss sharing arrangements, asset-backed financing, or other approved structures. Historically, the documentation and structuring overhead of Islamic finance products has made rapid digital scaling genuinely difficult. The theology and the technology have not always cooperated.

Work at the institutional level is starting to offer a template. Saudi Awwal Bank's execution of the first blockchain-based Islamic repurchase agreement in August 2025 — developed in partnership with digital-asset infrastructure provider Oumla — demonstrated that programmable Sharia compliance is achievable at the transactional level. By encoding compliance conditions directly into smart contracts and integrating with Chainlink's cross-chain infrastructure, SAB achieved T+0 settlement while eliminating the manual compliance verification steps that have historically slowed Islamic finance product deployment. Few outside specialist circles noticed. They should have.

Mal is attempting to apply a similar logic at the consumer and SME credit level. If Sharia-compliant credit structures can be automated and verified algorithmically, the operational cost of offering Islamic finance products at scale collapses. In markets where demand for compliant products runs high but supply has been throttled by complexity, that is a real commercial opportunity — not a theoretical one.

Risk Concentration and the Questions Investors Should Be Asking

The enthusiasm around AI credit scoring is well-founded. It is also incomplete. Sophisticated investors and family office principals need to look hard at the risks embedded in the model before writing cheques.

AI underwriting systems trained on historical data carry the risk of encoding existing biases or breaking down during economic conditions that differ materially from their training periods. A model calibrated on post-pandemic consumer behaviour in the UAE may perform very differently during a significant oil price correction or a regional liquidity tightening cycle. No one has stress-tested these systems through a genuine Gulf downturn. That gap deserves more attention than it currently receives.

Data privacy and regulatory arbitrage add further complications. As fintechs expand across multiple jurisdictions simultaneously — Mal is in licensing discussions across four countries — the consistency of their compliance frameworks will be tested. Investors, whether as direct equity holders or through venture allocations in family office portfolios, should examine not just the technology stack but the regulatory depth of the teams building it. Mal's leadership draws from Revolut and Nubank, which brings genuine operational scaling experience. But managing Islamic finance regulation across South and Southeast Asia simultaneously is a categorically different challenge from scaling a European neobank. Experience in one does not automatically transfer to the other.

The Forward View: Capital Allocation in an AI-Underwritten World

For private investors, family offices, and institutional allocators across the Gulf and beyond, the transformation of credit scoring by AI represents both a direct investment theme and a structural shift in how capital reaches portfolio companies. As AI-driven lending lowers the cost of credit assessment and expands the creditworthy universe, private credit strategies targeting underserved SME segments in markets like Egypt, Vietnam, and Nigeria become more viable and more scalable than they were even three years ago. The addressable market has not changed. The ability to profitably serve it has.

The $230 million that BlueFive Capital led into Mal reflects a conviction that AI-native financial infrastructure in Islamic markets is a multi-decade opportunity. That conviction is increasingly shared across the private capital community. The question for sophisticated allocators is not whether AI will reshape lending — it already has — but which platforms, in which regulatory environments, have built the data assets and compliance depth to convert the technology's promise into durable, risk-adjusted returns. That is a harder question. It is also the right one.