Banco Santander Opens Up Its AI Lab to the World with 11 Production-Grade GitHub Repositories
Banco Santander Opens Up Its AI Lab to the World with 11 Production-Grade GitHub Repositories #
Banco Santander released 11 internally developed artificial intelligence tools as open source on 25 June 2026, publishing them through a new GitHub organisation at github.com/SantanderAI under the Apache 2.0 licence.
The tools were built by the bank’s in-house AI Lab for use in financial operations. The repositories cover machine learning, large language models, generative AI, responsible AI, and AI governance. Each includes technical documentation, contribution guides, a code of conduct, and security policies.
Among the published tools is gen-fraud-graph, a synthetic data generator that can produce networks of fake transactions at scales exceeding 100 million nodes, allowing institutions to train fraud-detection models without using real customer data. Other repositories include autoguardrails, an alignment research scaffold for large language model guardrails; mech-gov-framework, a model-agnostic governance layer for high-stakes LLM decision systems; and mutatis-mutandis, a research tool for discrimination analysis and algorithmic fairness.
Financial institutions have historically treated algorithmic models as proprietary. Under the Apache 2.0 licence, any developer, researcher, or institution can use and modify the tools at no cost. Outside contributors may also submit pull requests to propose changes, making the release an ongoing collaborative project rather than a one-time publication.
Santander said the release followed internal reviews covering intellectual property, data protection, cybersecurity, and licensing. The bank described the move as part of its commitment to responsible innovation and cooperation with the technology community.
Copy link