The OECD.AI Policy Navigator

Our policy navigator is a living repository from more than 80 jurisdictions and organisations. Use the filters to browse initiatives and find what you are looking for.

Judicial Branch AI Management and Governance Model (Subnational: Mato Grosso)


-
Added by:   OECD analyst
Added on:   06 Oct 2026
Updated by:   OECD analyst
Updated on:   06 Oct 2026

The Judicial Branch (PJMT) adopts an AI governance model structured through strategic and operational committees responsible for guiding, prioritising, and overseeing institutional initiatives. The AI Center serves as the central hub, integrating the development, management, and application of AI across the judiciary. The model includes feasibility analysis, approval workflows, and budgetary, ethical, and legal controls, aligned with National Council of Justice (CNJ) Resolution No. 615/2025.

Name in original language

Modelo de Gestão e Governança de IA do Poder Judiciário de Mato Grosso (PJMT)

Initiative overview

The PJMT’s AI governance initiative was created to address the growing need for the structured, transparent, and accountable use of artificial intelligence within the Judiciary. As AI adoption expanded across different units, the absence of standardised processes posed risks to legal compliance, ethical use, resource allocation, and decision consistency. This initiative seeks to transform AI from isolated experiments into a coordinated institutional capability aligned with public value and regulatory requirements.

The main objective is to establish a comprehensive governance model that ensures all AI initiatives follow defined processes for feasibility assessment, prioritisation, approval, development, and monitoring. This includes the establishment of strategic and operational committees responsible for decision-making and oversight, as well as the strengthening of the AI Center as a central unit that integrates technical development with governance practices. The model is aligned with CNJ Resolution No. 615/2025, ensuring adherence to legal, ethical, and accountability standards.

The initiative currently encompasses a portfolio of ten AI projects under governance management, with six solutions in production and four in testing and development phases. In parallel, the AI Center has been proactively developing governance mechanisms to monitor key technical aspects such as token consumption, Retrieval-Augmented Generation (RAG) processes, and the management of data sources, ensuring efficiency, traceability, and responsible use of computational resources.

User training is a mandatory prerequisite for AI adoption, reinforcing the importance of human oversight and responsible interaction with AI systems. This approach ensures that users are properly prepared to understand the limitations, risks, and appropriate use of the technology within judicial processes.

Regarding implementation, the initiative is being institutionalised through formal governance structures, standardised workflows, and clear role definitions across the organisation. It is designed to scale by enabling multiple units to propose and develop AI solutions within a controlled framework, ensuring sustainability, interoperability, and continuous evolution of AI capabilities within the Judiciary.

Other relevant details

Results, outcomes and impacts: The initiative has already delivered significant results, particularly through LexIA, one of its main AI solutions in production. In its first ten days of use, LexIA increased productivity by over 100% in adopting units and reduced the time required to produce draft decisions by 42%. The platform currently has over 1,500 registered users and processes an average of 6,000 requests per day. Results are monitored through operational indicators, usage metrics, and performance tracking across projects. The initiative is expected to scale further, expanding AI adoption while strengthening governance, efficiency, and decision consistency across the Judiciary. Challenges and lessons learned: The initiative has faced challenges related to handling sensitive data, judicial confidentiality, and ensuring ethical and legal compliance in AI use. Additional complexities include managing multiple AI models within a single solution, controlling financial costs, such as token consumption, and mitigating risks arising from misuse or unintended outputs. These challenges have been addressed through governance processes, access controls, monitoring, and harm-containment protocols. A key lesson learned is the importance of user training and human oversight in responsible AI use. Success depends on strong senior leadership support, institutionalised governance, and alignment between technology, legal frameworks, and business areas, enabling the identification of high-impact AI opportunities.