Specialized Large Language Models for European Financial Compliance Automation

RegLLM aims to develop specialised Large Language Models (LLMs) trained on European financial regulatory data to automate and simplify compliance processes for financial institutions. The project addresses a key challenge faced by SMEs and regulated organisations, which still rely heavily on manual interpretation, spreadsheets, and fragmented tools to manage complex regulatory obligations under frameworks such as AML, PSD2, DORA, MiCA, MiFID II, EMIR, and Basel III/CRR. As part of the experiment, Complok will collect, preprocess, and structure large volumes of EU regulatory texts, then leverage EuroHPC infrastructure to fine-tune open-source LLMs for domain-specific compliance applications. The resulting RegLLM API will be integrated into Complok’s existing compliance platform, enabling organisations to access advanced regulatory intelligence directly within their workflows. The goal is to support faster and more consistent interpretation of regulatory requirements, automate compliance gap analysis, provide predictive risk alerts, and deliver transparent and auditable compliance assistance for fintech companies and other regulated institutions. By reducing manual effort and improving regulatory oversight, RegLLM aims to make compliance management more efficient, scalable, and accessible.

Organisations involved: 

End-User: Complok