From Fragmentation to Cohesion: Multilingual Legal AI for Europe's Evolving Data Landscape

Abstract - The growing complexity of legal frameworks like General Data Protection Regulation (GDPR) and Artificial Intelligence (AI) Act, with multilingual and jurisdictional variations, makes regulatory compliance difficult. General-purpose large language models (LLMs) struggle with domain-specific interpretation, frequently producing hallucinated and misleading outputs. To address this, we introduce LexAid-EU, a multilingual, jurisdiction-aware AI framework combining retrieval-augmented generation (RAG) with LLMs. Legal texts are segmented, embedded, and stored in a vector database for semantic search and contextual response generation. The system reduces hallucination by grounding answers in authoritative legal sources. A comprehensive evaluation of context awareness, bias & hallucination, robustness, and language support was conducted using established metrics. LexAid-EU, tested with open-source and paid LLMs, shows substantial improvements in legal reasoning accuracy and reliability across languages and jurisdictions. LexAid-EU offers a scalable, explainable, compliant solution supporting legal professionals in Europe’s evolving data governance landscape.