Machine learning-based credit scoring is reshaping the UK lending landscape, with alternative data models enabling financial institutions to extend credit to an estimated 2.4 million consumers who were previously excluded by conventional scoring systems. Companies including ClearScore, Credit Kudos (now part of Apple), and Experian's new ML-powered platform are using data points such as rent payment history, utility bills, open banking transaction patterns, and even employment contract data to build more nuanced risk profiles. Early results suggest that these models can identify creditworthy borrowers among traditionally underserved populations — including gig economy workers, recent immigrants, and young adults with thin credit files — without increasing default rates.

"Traditional credit scoring is essentially a blunt instrument that penalises people for being invisible to the financial system," said Freddy Kelly, founder of Credit Kudos. "Our models look at actual financial behaviour — how someone manages their money day to day — rather than relying on a narrow set of historical credit data." Monzo and Starling Bank have been early adopters, using ML-based assessments to underwrite personal loans and overdrafts for customers who would be automatically rejected by legacy systems. Monzo reported that its ML-scored lending cohort had a 90-day arrears rate of just 1.8 per cent, compared with 2.1 per cent for its conventionally scored portfolio.

The trend has attracted both praise and scrutiny. The Treasury's Financial Inclusion Commission welcomed the technology's potential to address the estimated 5.8 million UK adults who are either unbanked or underserved by mainstream financial services. However, the Equality and Human Rights Commission has called for rigorous testing to ensure that ML models do not inadvertently discriminate on the basis of protected characteristics such as race, gender, or disability. "The opacity of machine learning models makes it harder to detect bias than in traditional scoring systems," said Rebecca Hilsenrath, the Commission's chief executive. The FCA's forthcoming AI regulation framework is expected to include specific provisions for algorithmic credit decisioning, including mandatory fairness audits and adverse action explanations.