Regulatory compliance in Europe is increasingly shaped by large, overlapping bodies of regulation such as the EU AI Act, Data Act, and sector-specific safety and data governance regimes. This project examines how neuro-symbolic AI can be used to support explainable, auditable, and scalable regulatory compliance across complex organizational and ecosystem contexts.

The project is developed through an applied research pilot, ComplianceTwin, conducted in collaboration with five Finnish enterprises operating in highly regulated and data-intensive domains. Participants include Tieto (digital services and cloud infrastructure), Vaisala (industrial measurement and environmental sensing), iLOQ (secure digital access systems), PwC Finland (professional services and regulatory advisory), and one large industrial manufacturer operating global product portfolios. Together, these organizations represent diverse compliance environments spanning software, IoT, industrial products, data services, and professional advisory.

The research investigates how regulatory texts, like statutes, standards, and guidance, can be transformed into structured legal knowledge by capturing obligations, exceptions, scope, and hierarchy. The focus is to support applicability analysis, obligation mapping, gap identification, and evidence generation in a manner that remains traceable and defensible in audits and enforcement.

The pilot is structured as a seven-month program (November 2025–May 2026) consisting of four collaborative workshops and ongoing between-session work. The phases include: (1) ecosystem mapping and identification of cross-organizational compliance challenges; (2) translation of regulatory requirements into software logic and AI agentic workflows; (3) prototype development and testing; and (4) deployment and evaluation.

Key deliverables include a technical and governance blueprint, a prototype compliance twin, and an evaluation report documenting accuracy, efficiency, and trust considerations. The project contributes to research on legal knowledge representation, computable law, and explainable AI, while offering empirical insights into how AI-assisted legal reasoning can support, rather than replace, expert judgment in high-stakes regulatory settings.