Santander opens 11 AI projects to developers on GitHub

Banco Santander has published details of 11 artificial intelligence projects on a new GitHub channel, making selected AI Lab work available under an open-source licence. The move adds a public technical cooperation channel to the Spanish bank’s broader AI programme, which Santander has said it expects to generate more than €200 million in business value in 2026.

The bank said the GitHub release is intended to support technical cooperation and shared learning. Santander described the published work as tools, examples and resources rather than systems containing real customer information.

José Manuel de la Chica, head of Santander AI Lab, said the next stage of artificial intelligence would depend not only on access to advanced models, but also on the capacity to use them with “rigour, confidence and responsibility.” In banking, he said, that requires evidence that systems are “secure, fair, robust and auditable.”

The GitHub channel is designed to give external developers, researchers and other technical users access to selected components created inside Santander AI Lab. Open-source publication allows users to inspect code, test methods and adapt the material within the terms of the licence, while the bank retains control over what it chooses to release.

One of the projects Santander has made public is gen-fraud-graph, a tool for creating synthetic networks of fraud-related transactions and behaviours. Santander said the tool is intended to help identify complex fraud patterns without exposing people’s private information.

Synthetic data is used in AI development when organisations want to test models or analytical approaches without relying on real customer records. In this case, Santander said the tool produces artificial transaction and behaviour networks linked to fraud analysis, giving users a way to study pattern detection while limiting privacy risks.

Another released project is mutatis-mutandis, which Santander said addresses fairness in AI systems. The project focuses on the challenge of ensuring that people in comparable situations are treated equally by algorithmic systems.

Santander said mutatis-mutandis examines algorithmic discrimination through counterfactual comparators. That approach tests how an AI system treats comparable cases, helping analysts assess whether different outcomes are associated with characteristics that should not drive a decision.

The publication follows Santander’s recent plan to extend AI access to all of its 185,000 employees. The bank has said the technology is beginning to produce tangible results across the business and has tied its internal rollout to the expectation of more than €200 million in business value in 2026.

Santander’s GitHub channel is available at github.com/SantanderAI.

This story draws on original reporting from Finextra Research.