AI in Financial Services: MENA Perspectives

The transformation of Middle Eastern financial services through artificial intelligence represents one of the most significant shifts in the region’s economic history. Banks that operated on largely unchanged principles for decades are now racing to integrate AI across their operations, from customer service chatbots handling routine enquiries to sophisticated fraud detection systems processing millions of transactions in real time. The Gulf states have emerged as particularly aggressive adopters, with UAE banks investing an estimated $500 million in AI initiatives in 2023 alone according to PwC Middle East analysis. This investment reflects recognition that AI is not merely an operational improvement but a competitive imperative—that institutions which fail to develop AI capabilities risk losing market share to more sophisticated rivals and eventually to technology companies entering financial services from adjacent industries.

The regulatory environment has evolved to encourage innovation while maintaining stability concerns that prudential authorities understandably prioritise. The Central Bank of the UAE has established its AI and Digital Transformation Office, developing frameworks that enable experimentation while protecting consumers and systemic stability. Saudi Arabia’s central bank has launched regulatory sandbox programmes that allow fintech companies to test AI applications under controlled conditions. Bahrain’s Central Bank has positioned itself as a regional hub for financial technology, developing regulatory frameworks specifically designed to accommodate AI-driven services. These regulatory approaches reflect a broader regional recognition that financial services innovation—including AI—requires regulatory frameworks that enable rather than merely restrict. The challenge lies in developing rules sophisticated enough to address AI’s unique characteristics, including opacity, autonomy, and continuous learning that may produce unexpected behaviours.

Customer expectations have driven much of the AI adoption pressure, particularly among younger demographics who expect digital experiences comparable to global technology leaders. A regional survey by Accenture found that over 70% of MENA banking customers under 35 would consider switching to a digital-only bank or fintech provider, citing better user experience as a primary motivation. This generational shift poses existential questions for traditional institutions: can they transform quickly enough to retain customers whose expectations are shaped by experiences with Amazon, Apple, and Google? AI capabilities have become central to this competitive dynamic, enabling the personalised recommendations, instant responses, and seamless experiences that younger customers expect. Banks that cannot deliver AI-enabled services risk being perceived as outdated institutions that younger customers tolerate rather than choose—a perception with serious implications for long-term market position and profitability.

Credit, Risk, and the Algorithmic Assessment of Borrowers

AI-enabled credit assessment represents perhaps the most consequential application of artificial intelligence in MENA financial services, with implications extending far beyond operational efficiency to fundamental questions about access to capital and economic opportunity. Traditional credit scoring in the region has relied heavily on employment history, salary, and existing banking relationships—criteria that systematically disadvantage informal workers, entrepreneurs, women with interrupted career histories, and young people without established credit records. AI systems can potentially incorporate alternative data sources—mobile phone usage patterns, utility payment histories, social media activity—to assess creditworthiness among populations that traditional models cannot evaluate. The promise is financial inclusion at scale: extending credit to segments previously excluded from formal financial services. The risk is that algorithmic systems may encode existing biases in new forms, making discrimination more systematic and harder to identify than traditional underwriting prejudices.

The evidence on AI credit scoring’s impact on fairness remains contested and context-dependent. Research published through SSRN has documented cases where machine learning models reduced racial disparities in credit decisions compared to traditional models, suggesting that algorithmic systems can outperform human judgement on fairness dimensions. However, other studies have found the opposite—that AI systems trained on historical data perpetuate and amplify historical patterns of discrimination. A National Bureau of Economic Research study examining algorithmic lending in the United States found that minority borrowers paid higher interest rates than equally qualified non-minority borrowers, with the gap persisting across algorithmic and human-based lending. For MENA applications, these findings suggest that AI credit scoring’s fairness impact depends critically on how systems are designed, what data they incorporate, and how institutions monitor and address discriminatory patterns. Absent deliberate attention to fairness, efficiency gains may come at equity costs that regulators and society should not accept.

Regional financial institutions have begun deploying AI credit assessment despite these unresolved questions, driven by competitive pressure and the genuine potential for financial inclusion advancement. Emirates NBD has implemented AI-driven credit decisioning for consumer lending, reporting significant reductions in processing time and improved risk prediction accuracy. Mashreq Bank has developed AI models incorporating alternative data for credit assessment among customers with limited traditional credit histories. Saudi Arabia’s fintech sector includes several companies—Tamara, Tabby, and others—using AI for buy-now-pay-later credit decisions made in seconds. These deployments proceed ahead of comprehensive regulatory frameworks for algorithmic fairness, raising questions about consumer protection that regulators are racing to address. The UAE Central Bank has issued guidance on AI use in financial services that addresses some fairness concerns, while Saudi Arabia’s central bank has incorporated algorithmic accountability principles into fintech licensing requirements. Whether these frameworks prove adequate to protect consumers while enabling innovation remains to be demonstrated.

Fraud Detection and the Endless Arms Race

Financial fraud in the MENA region has evolved dramatically with digitalisation, creating both urgent need and compelling use cases for AI-enabled detection. The shift to digital payments—accelerated by pandemic-era behaviour changes and government initiatives to reduce cash dependency—has expanded the attack surface available to fraudsters. Morgan Stanley research estimates that digital payment volumes in the Middle East will grow at 20% annually through 2026, representing over $2 trillion in transactions. Each transaction represents a potential fraud vector; the scale makes human review impossible and rule-based systems inadequate to detect sophisticated attacks. AI offers the pattern recognition capability necessary to identify anomalies across vast transaction volumes—finding the fraudulent needle in the legitimate haystack before losses occur. The technology has moved from experimental to essential, with major MENA banks reporting that AI systems now handle the vast majority of fraud detection decisions with human review reserved for edge cases.

The arms race dynamic between fraud detection and fraud execution defines AI’s role in financial security. Fraudsters continuously probe systems for vulnerabilities, adjusting tactics when detection improves. AI enables detection systems to adapt continuously as well, learning from new fraud patterns and adjusting models without requiring human analysts to identify and program responses to each new tactic. Mastercard reports that its AI-enabled fraud detection systems evaluate over 1 billion transactions daily, making decisions in milliseconds that balance fraud prevention against false positives that frustrate legitimate customers. In the MENA region, where cross-border transactions are common and fraudsters exploit jurisdictional complexity, AI systems must additionally navigate the challenge of identifying legitimate but unusual transaction patterns—the businessman making a large purchase in a new country, the expatriate sending remittances home through unfamiliar channels. These systems increasingly incorporate contextual intelligence: understanding that a transaction pattern unusual for one customer segment may be perfectly normal for another, adjusting sensitivity accordingly.

Despite AI’s capabilities, fraud losses continue to grow across MENA financial services, reflecting both increasing digital transaction volumes and fraudster sophistication that keeps pace with detection improvements. Social engineering attacks—manipulating customers into authorising fraudulent transactions themselves—have proven particularly difficult for AI to address, since the transactions appear legitimate from the bank’s perspective until the customer reports being deceived. Synthetic identity fraud, where criminals create fictitious identities combining real and fabricated information, challenges AI systems trained to identify fraudulent use of real identities. LexisNexis Risk Solutions research indicates that every dollar of fraud loss costs financial institutions nearly four dollars when including investigation, recovery, and compliance expenses. These economics drive continued investment in AI detection capabilities while also highlighting the limits of technological solutions—fraud is ultimately a human problem that technology can address but not solve entirely. The most effective approaches combine AI detection with customer education, strong authentication protocols, and response capabilities that minimise losses when fraud does occur.

The Automation of Advice and the Future of Wealth Management

Wealth management in the MENA region has traditionally served high-net-worth individuals through relationship-based advisory that combines investment expertise with personal service. The economics of this model—human advisors who cost substantial amounts to employ—have necessarily limited access to professional financial advice, leaving mass-market customers to navigate investment decisions without guidance or to accept standardised products that may not suit their circumstances. AI-enabled robo-advisory offers potential to democratise wealth management by automating the analysis and recommendation functions that previously required expensive human expertise. Statista forecasts suggest that robo-advisory assets under management in the Middle East will exceed $5 billion by 2027, representing a small but rapidly growing segment of the regional wealth management market.

Regional wealth managers have responded to the robo-advisory trend through hybrid models that combine algorithmic efficiency with human relationship management. Emirates NBD’s Liv platform incorporates AI-driven investment recommendations alongside traditional banking services, targeting younger customers comfortable with digital wealth management. Saudi Arabia’s Wahed Invest has built a Shariah-compliant robo-advisory platform serving customers across the Muslim world, using AI to construct portfolios that satisfy religious requirements while optimising returns. Sarwa, based in the UAE, has attracted significant investment for its AI-enabled wealth management platform targeting expatriates and younger local customers. These platforms must navigate the challenge of earning trust for high-stakes financial decisions: customers comfortable using AI for restaurant recommendations may hesitate to delegate retirement planning to algorithms. Effective robo-advisory platforms address this trust gap through transparency about how recommendations are generated, human support available when customers want it, and track records demonstrating competent performance through market cycles.

The competitive dynamics unleashed by AI in wealth management extend beyond robo-advisors to reshape traditional advisory practices. Human advisors increasingly use AI tools to enhance their capabilities—generating investment analyses more quickly, monitoring client portfolios for rebalancing opportunities, identifying life events that might affect financial planning needs. This augmentation model may prove more durable than pure automation, combining AI efficiency with human judgement and relationship management that clients value. Oliver Wyman analysis suggests that the most successful wealth managers will be those who integrate AI capabilities most effectively into human-centred service models rather than choosing between human and algorithmic approaches. For MENA institutions, this insight suggests that AI investment should focus on augmenting advisor capabilities alongside developing standalone robo-advisory products—building AI as a tool that enhances human expertise rather than as a replacement that eliminates it. The future of wealth management likely involves both approaches serving different market segments, with AI enabling the expansion of professional financial advice to populations previously excluded by economics.