Argentina - Virtual Assistant Implementation (1486/2025)

Overview

Resolución AG 1486/2025 approves and launches a staged project to implement an AI-based Virtual Assistant integrated into the Poder Judicial de la Nación's Portal de Gestión de Causas (Lex100). The resolution identifies institutional actors, assigns technical and operational responsibilities, and establishes a controlled pilot for external users selected by the Federación Argentina de Colegios de Abogados (FACA). The tool's stated objective is to provide automated, natural-language responses to frequent user queries, improve user experience and increase operational efficiency while ensuring that unresolved queries are triaged to human operators. For official project announcements and pilot pre‑registration see Consejo de la Magistratura announcement (03/09/2025) and the pre-registration portal at PJN services - Preinscripción. ([consejomagistratura.gov.ar](https://consejomagistratura.gov.ar/index.php/2025/09/03/avanza-la-implementacion-de-un-asistente-virtual-con-inteligencia-artificial-en-el-poder-judicial/?utm_source=openai))

Definitions

This section provides operational definitions used throughout the resolution and the pilot materials: "Virtual Assistant" (the AI conversational interface integrated with the Portal de Gestión de Causas), "Knowledge Base" (validated, official content curated by the Departamento de Capacitación y Gestión de Desarrollo), "External User" (lawyers, peritos, public representatives interacting with Lex100), "Pilot" (controlled testing phase with selected participants), "Human Fallback" (automatic routing to SAUE human operators when the assistant cannot resolve a query), and "Processing API" (third‑party or hosted model inference interface used to execute natural language processing). These terms are consistent with the project's annex and technical glossary referenced in the formal approval.

Governance and Institutional Framework

The resolution places governance responsibility within the Consejo de la Magistratura and operational execution under the Dirección General de Tecnología, with the Departamento de Capacitación y Gestión de Desarrollo tasked with the knowledge base. The Administrador General of the PJN is identified as the signatory authority who approved the project. Oversight responsibilities include periodic reporting to the Consejo, coordination with FACA for participant selection, and collaboration with Dirección General de Seguridad Informática for technical security controls. The pilot governance model sets out clear escalation paths: unresolved outputs are diverted to SAUE; security incidents are reported to internal cybersecurity units; and a focal point is designated for external stakeholder communications. Institutional announcements summarising these roles are available at the Consejo information page. Consejo release and the PJN services area contain operational documentation and contact points used during the pilot. ([consejomagistratura.gov.ar](https://consejomagistratura.gov.ar/index.php/2025/09/03/avanza-la-implementacion-de-un-asistente-virtual-con-inteligencia-artificial-en-el-poder-judicial/?utm_source=openai))

Key Focus Areas

The resolution and associated pilot documentation emphasise a set of priority domains: (1) User transparency and notices (clear statement of scope and limitations of the assistant); (2) Human oversight and fallback to SAUE; (3) Data minimization and the exclusive use of curated official content for the knowledge base during phase one; (4) Logging, audit trails and metrics collection (conversations, intents not resolved, referral rates); (5) Security controls for API communication and credential management; and (6) Iterative evaluation and user feedback processes. Media and institutional reports note that the first phase purposefully restricts the assistant’s remit to avoid automated legal determinations and reduce risks to fundamental rights and due process. See the PJN pre‑registration documentation for participant obligations and confidentiality commitments. PJN - Preinscripción. ([servicios.pjn.gov.ar](https://servicios.pjn.gov.ar/formularios_IA/?utm_source=openai))

Implementation Framework

Implementation is framed as a two‑phase programme. Phase 1 (external beta) focuses on public-facing informational assistance and requires: curation of the knowledge base, participant selection and onboarding (FACA and provincial colleges), technical integration with Lex100, and a controlled access model with weekly cohorts. Phase 2 expands internal functionalities for PJN staff (administrative tools, supervised access to internal records) and contemplates additional safeguards such as on-premise models or institutionally hosted model instances to reduce third‑party dependency. The resolution's annexes (project specification and glossary) provide the technical boundaries, expected throughput, and projected token consumption metrics used to budget the pilot; journalistic reporting summarises these operational projections. Quórum report (28/05/2025). ([revistaquorum.com.ar](https://revistaquorum.com.ar/2025/05/28/crean-un-asistente-virtual-para-los-usuarios-del-poder-judicial-de-la-nacion/?utm_source=openai))

Monitoring and Evaluation

The pilot design requires active monitoring through logged conversation transcripts (subject to privacy controls), KPI dashboards (resolution rate, escalation rate, user satisfaction), and structured feedback forms completed by pilot participants. The Departamento de Capacitación will analyse metrics and produce periodic reports with recommendations for technical tuning, content updates, and risk mitigation. The Dirección General de Seguridad Informática will perform security audits and vulnerability assessments before any expansion beyond the pilot. Official communications indicate that training sessions and onboarding will precede pilot entry to reduce misuse and establish participant obligations. Consejo announcement. ([consejomagistratura.gov.ar](https://consejomagistratura.gov.ar/index.php/2025/09/03/avanza-la-implementacion-de-un-asistente-virtual-con-inteligencia-artificial-en-el-poder-judicial/?utm_source=openai))

Penalties, Liability, and Appeals

While the summary materials released with the resolution do not list monetary fines, they preserve the application of existing disciplinary and administrative regimes for judicial staff and established contractual remedies for external vendors. The pilot imposes participant obligations (confidentiality, no public dissemination of test materials), and breach of those obligations can result in removal from the programme and referral to disciplinary authorities. The resolution instructs that incidents affecting personal data, system security or rights of users must be reported and investigated, and that corrective actions will be taken proportionate to the harm and institutional rules. Applicable liability for third‑party providers will be managed through procurement contracts and service level agreements (SLAs) containing indemnities and security obligations.

Relationship to Other Instruments

Resolución 1486/2025 is situated within a broader public‑sector trend to modernise judicial administration and to pilot AI tools under controlled conditions. It interacts with existing PJN internal regulations on IT security, public administration procurement rules, and applicable data protection obligations under national law and agency policies. The resolution references technical annexes and project documentation that align with institutional standards for software deployment, vendor engagement, and training. Public statements by the Consejo and PJN form part of the record that contextualises regulatory oversight and the staged approach. Justice Ministry summary. ([justicia.ar](https://justicia.ar/novedades/1661/2025/09/avanza-la-implementacion-de-un-asistente-virtual-con-inteligencia-artificial-en-el-poder-judicial?utm_source=openai))

International Alignment

The initiative references international best practices in AI deployment for the public sector—prioritising transparency, human oversight and data minimisation—and is consistent with emerging OECD and G7 guidance encouraging pilot testing, risk assessment and iterative deployment of AI in government services. Media and policy observers have noted similarities between this approach and international frameworks that separate low-risk informational assistants from high-risk automated decision systems, and that recommend progressive expansion only after validated safety and privacy controls are in place. For international context see policy analyses and commentary summarising Argentina’s institutional pilots. IAPP coverage.

Implementation Timeline

Sources and References