Akoma Ntoso: Powering Legal AI with Open Data Standard

In the era of Artificial Intelligence, the utilization of accurate and authoritative legal open data is essential for integrating normative policies within eGov services, embedding the principles of the Rule of Law into Large Language Models (LLMs) and Digital Twin simulations, forecasting future regulatory-compliant scenarios, supporting decision-makers, and managing societal complexity through legal norms. Furthermore, such data plays a pivotal role in enhancing eParticipation and enriching democratic discourse. Recent developments in Generative AI (GenAI) and LLMs in the legal domain have highlighted significant challenges related to the FAIR principles—Findability, Accessibility, Interoperability, and Reusability. In particular, it is critical to ensure reproducibility of outcomes, accessibility of AI models, explainability of results, and availability of original data sources. Legal open standards, such as Akoma Ntoso, are instrumental in addressing these challenges and mitigating the inherent risks associated with AI applications in the legal domain. Widely adopted by international organizations, including the United Nations, as well as by national parliaments and official gazettes, Akoma Ntoso serves as a foundational pillar for the development of legal AI applications (e.g., GENAI4LEX of UNIBO). This talk presents the core features of Akoma Ntoso (the LegalDocML standard endorsed by OASIS), demonstrating how its semantic annotation facilitates the unlocking of legal knowledge and its application in advanced legal AI systems.