The study on ‘Proof of Context of Context Aware Legal Verification’ is a third small scale study that is funded by the Commission and in which we explore ‘Legal drafting in the age of AI and digitisation’, a journey that we started at the end of 2020. 

The previous studies can be found here Drafting legislation in the era of AI and digitisation | Interoperable Europe Portal and here AI-based solutions for legislative drafting in the EU - Publications Office of the EU

This third study is carried out by the University of Bologna and is about demonstrating the feasibility of implementing smart functionalities that can assist a user to draft high-quality legislation. The study is about turning an ‘XML editor’ into a ‘smart legal drafting solution’.

Starting point

The study builds on two main pillars: 

• LEOS and AKN4EU - for syntax and format. LEOS is a web tool to support the legislative drafting in the EU. The version we launched in mid-2025, is a robust legal editing solution that supports amongst others collaborative editing, document management, fine grained change tracking and version handling. Moreover, and most importantly, it takes care of structure and produces machine readable law in AKN4EU. EdiT, the Commission instantiation of LEOS, is currently in use in the Commission.
• Smart functionalities – semantics and meaning. The study builds on the concept of smart functionalities, i.e. advanced and intelligent software capability to enhance the LEOS system (see more here). Smart functionalities use both AI and ‘legacy’ technologies in essence to ‘codify legal meaning’, a most difficult task requiring an in-depth understanding of legal theory and practice and a realistic appreciation of the potential use of AI. 

The study more specifically zooms in on the topic of using hybrid AI to implement context-aware legal verification. 

Study results – An overview

The study (i) piloted the use of AI technologies to verify ‘in the specific context of an act being drafted’ legal references and normative definitions, and (ii)  used LEOS as a platform for integration of services such as handling regulatory reporting requirements (see Streamlining regulatory reporting: the SORTIS project results | Interoperable Europe Portal ). Moreover,  the study developed a framework architecture capable of enhancing LEOS with add-ons, developed using AI technologies. 

Why it matters – Proof of Value

The Proof of Concept (PoC) convincingly demonstrated that it is feasible to implement smart functionalities in a relatively short time using limited resources relying on readily available AI technologies and existing data sets. 

It also successfully showed that LEOS is an excellent integration platform to host a rich variety of smart functionalities and thus provides a flexible and powerful legal drafting solution. 

It proved that the implementation of smart functionalities has a high potential for improving the quality of legislation and increase productivity of law-making, two elements of high importance in view of the current developments. (see e.g., EU competitiveness - European Commission)

And, enabling machine readable law (i) allows for seamless cooperation between the many actors involved in drafting legislation, (ii) makes law accessible and easily searchable, and (iii) provides an opportunity to use the best of technology. 

Deeper dive in the results

The architecture is depicted below. The overall architecture is composed of an XML database that includes the Akoma Ntoso XML documents and an SQL database containing the correspondence between each document and its EUROVOC categorisation.

The three use cases are as follows:

• Context aware verification of legal definitions. Legal definitions are an essential part of law. The PoC implemented functionality to assist a law-drafter/policy developer in a consistent and relevant use of legal definitions. The system can, considering the content of the act being drafted, provide a link to or the text of relevant legal definitions. Legal definitions are ranked by relevance. However, it is up to the user to make  an informed decision on which definition to use. Additionally, and equally important, the PoC provides information on the time validity of the legal definition.
• Context aware verification of normative references. Normative references are qualified citations to other documents or provisions that are relevant for the normative discourse. The PoC is capable to complete incomplete normative references and to retrieve and rank ‘highly relevant’ existing and ‘into force’ references .
• Supporting reporting requests in LEOS. The SORTIS (Study on Regulatory Reporting Standards) project aimed to develop a solution for detecting a reporting request (RR) inside European legislation and defined a regulatory reporting metadata vocabulary (RRMV). This PoC integrated this functionality in LEOS (i) to qualify during the drafting a provision as “RR”, (ii) to use AI to automatically detect the components of the RR, (iii) to extract the parameters in Resource Description Framework (RDF), and (iv) to clearly display the information to allow the user to verify the correctness of the RR.

References and other relevant resources

Legal drafting is an essential task of any government. Machine readable law is transforming how laws are created. The responsible and smart use of IT and AI is a main vehicle to improve the quality, efficiency, and transparency of laws and the law-making process. See also

• Digital-ready policy making in which digital-ready drafting is a main component (see Digital-ready policies | Interoperable Europe Portal),
• European Commission digital strategy Next generation digital Commission (see  European Commission Digital Strategy - European Commission) which sets out the way for the Commission to become a more agile, trusted and proactive administration supporting the delivery of EU’s strategic priorities.

Want to know more?

See the attached full study and follow us for more updates.