Argentina - San Juan - AI Judicial Protocol (2024)

Overview

The San Juan AI Judicial Protocol represents a pioneering regulatory effort by the Court of Justice of San Juan, Argentina, to institutionalize the use of Artificial Intelligence (AI) within the provincial judiciary. Recognizing the transformative potential of digital technologies, the protocol establishes a clear legal and ethical boundary for the integration of automated systems in judicial processes. The primary objective is to enhance the efficiency of the justice system—reducing backlogs and accelerating procedural timelines—while strictly adhering to the constitutional guarantees of due process, equality before the law, and judicial independence. This regulation is not merely a technical guide but a binding framework that governs how judicial officials interact with AI tools, ensuring that technology serves as an auxiliary instrument rather than a replacement for human judgment. The protocol was formally adopted through Acuerdo General N° 124/2024, marking a significant milestone in the digitalization of the Argentine legal landscape. It addresses the growing need for a structured approach to innovation, ensuring that the adoption of AI does not compromise the core values of the judicial function. By providing a centralized set of rules, the Court of Justice aims to foster a culture of responsible innovation, where technology is used to empower judicial officers and improve the quality of service provided to the citizens of San Juan.

Definitions

The protocol provides precise definitions to ensure legal certainty across the judicial branch. It defines Artificial Intelligence as a set of technologies capable of performing tasks that typically require human intelligence, such as pattern recognition, natural language processing, and predictive analysis. A critical distinction is made between 'Generative AI,' which creates new content based on training data, and 'Predictive AI,' which analyzes historical data to suggest outcomes or trends. The document emphasizes the concept of 'Human-in-the-Loop,' a mandatory requirement where a human official must review, validate, and take responsibility for any output generated by an AI system before it gains legal effect. Furthermore, the protocol defines 'Algorithmic Transparency' as the ability of the judiciary to explain the logic, data inputs, and intended purposes of any AI tool used in a case. It also introduces the term 'Judicial Automation' to describe the use of software to handle repetitive, low-discretion tasks, such as scheduling or basic document categorization. These definitions are crucial for establishing a common technical and legal language, allowing for better communication between IT departments and judicial chambers. By clarifying these terms, the protocol ensures that all stakeholders understand the scope and limitations of the technologies being deployed, thereby reducing the risk of procedural errors or misinterpretations of AI-generated outputs.

Governance and Institutional Framework

The governance of AI within the San Juan judiciary is centralized under the Court of Justice (Corte de Justicia de San Juan), which acts as the ultimate regulatory and oversight body. The protocol delegates technical implementation and daily management to the Secretariat of Informatics (Secretaría de Informática) and the Planning Department. These bodies are responsible for selecting, testing, and deploying AI tools, ensuring they meet the security and ethical standards defined in the protocol. This institutional structure is designed to prevent the fragmented or unauthorized use of AI tools by individual courts, mandating that any AI system used must be officially vetted and approved by the central administration. In addition to technical oversight, the framework establishes a multidisciplinary approach to governance. It encourages the formation of advisory committees comprising legal experts, ethicists, and computer scientists to periodically review the impact of AI on judicial outcomes. This governance model emphasizes accountability; the Court of Justice maintains the power to suspend the use of any AI tool if it is found to produce biased results or compromise the integrity of the judicial process. By centralizing control, the protocol ensures that the adoption of AI is consistent across different jurisdictions within the province, maintaining the principle of legal unity and preventing 'digital divides' between different courts. The framework also includes mechanisms for inter-institutional cooperation, allowing the San Juan judiciary to collaborate with other provincial courts and national bodies to share best practices and technical resources.

Key Focus Areas

The protocol focuses heavily on the ethical deployment of AI, with a particular emphasis on the prevention of bias and discrimination. It mandates that all AI systems must be trained on high-quality, representative data sets to avoid replicating historical prejudices against marginalized groups. Another key focus area is 'Data Protection and Privacy.' Given that the judiciary handles sensitive personal information, the protocol requires that AI tools implement 'privacy by design' and 'privacy by default' principles. This includes the anonymization of personal data when used for training AI models and the implementation of robust cybersecurity measures to prevent data breaches or unauthorized access to judicial files. Another significant focus is the 'Explainability' of AI-assisted decisions. The protocol asserts that litigants have the right to know if an AI tool was used in their case and to understand the criteria used by the algorithm. This is essential for upholding the right to a reasoned and justified judicial decision (sentencia fundada). The document also highlights the importance of 'Human Oversight,' stating that AI should never be used to automate the core 'decisional' aspect of a judge's work in complex cases involving high levels of discretion or fundamental rights. Instead, AI is prioritized for 'auxiliary' tasks, such as summarizing case files, searching for relevant precedents, or managing administrative workflows. The protocol also addresses the need for 'Technical Robustness,' ensuring that AI systems are reliable, accurate, and resilient to adversarial attacks or technical failures.

Implementation Framework

The implementation of the San Juan AI Judicial Protocol follows a phased and iterative approach. The first phase involves the identification of 'pilot projects' where AI can provide immediate value with low risk, such as the automation of notification systems or the classification of incoming lawsuits. Before any AI tool is fully integrated, it must undergo a rigorous 'Impact Assessment' that evaluates its potential effects on human rights, technical reliability, and procedural efficiency. This assessment must be documented and made available for internal review, ensuring that the transition to AI-assisted justice is evidence-based and transparent. To support this implementation, the protocol mandates continuous training for judicial personnel. This training is not limited to technical usage but includes 'AI Literacy,' covering the ethical risks, limitations, and legal implications of the technology. The Court of Justice is tasked with developing educational programs in collaboration with universities and research centers. Furthermore, the implementation framework includes a 'Technical Standards' section, which specifies the requirements for interoperability between AI tools and existing judicial management software. This ensures that AI integration does not create technical silos but rather enhances the existing digital infrastructure of the San Juan courts. The protocol also establishes a feedback loop where judicial officers can report issues or suggest improvements to the AI tools, ensuring that the technology evolves in response to the practical needs of the courtroom.

Monitoring and Evaluation

Monitoring and evaluation are integral to the protocol's long-term success. The Court of Justice is required to conduct periodic audits of all AI systems in use to ensure they remain compliant with the ethical and technical standards set forth in the regulation. These audits focus on 'Algorithmic Drift'—the phenomenon where an AI model's performance degrades or changes over time as it encounters new data—and 'Bias Detection.' If an audit reveals significant deviations or unfair outcomes, the protocol provides a mechanism for the immediate decommissioning or recalibration of the offending system. Evaluation also involves gathering feedback from the end-users: judges, lawyers, and the public. The protocol suggests the use of performance indicators (KPIs) to measure the impact of AI on the speed of case resolution and the quality of judicial service. These metrics are intended to provide a transparent view of the 'Return on Investment' for AI technologies, not just in financial terms but in terms of 'Social Value' and 'Access to Justice.' Annual reports summarizing the findings of these monitoring activities must be published, fostering a culture of transparency and allowing for public scrutiny of how technology is being used to administer justice in the province. The monitoring process also includes a review of the technical infrastructure to ensure that the hardware and software supporting AI systems are up-to-date and secure against emerging cyber threats.

Penalties, Liability, and Appeals

The protocol clarifies the lines of legal liability in the age of AI. It explicitly states that the use of an AI tool does not exempt a judicial official from their legal responsibilities. If a judge relies on an AI-generated draft that contains errors or violates the law, the judge remains personally and professionally liable for the final decision. This reinforces the 'Human-in-the-Loop' principle, ensuring that AI is treated as a sophisticated assistant rather than a surrogate decision-maker. The document also outlines that any technical failure of an AI system that results in a violation of procedural rights can be grounds for an appeal or a declaration of nullity of the affected judicial act. Regarding administrative penalties, the protocol stipulates that the unauthorized or improper use of AI tools—such as using non-approved external generative AI for sensitive case work—may lead to disciplinary action against judicial staff. This is particularly relevant for the protection of confidentiality; using public AI platforms to process private case data is strictly prohibited. The protocol also ensures that litigants have a clear 'Right to Appeal' any decision where they believe an AI tool was used improperly or where the algorithm produced a biased result. This right to appeal is a critical safeguard, ensuring that the introduction of technology does not diminish the traditional protections of the adversarial legal system. The protocol also establishes that the judiciary must maintain a log of AI usage to facilitate investigations into potential misconduct or technical errors.

Relationship to Other Instruments

The San Juan AI Judicial Protocol operates within a broader legal ecosystem. It is designed to complement the Argentine National Constitution and the Provincial Constitution of San Juan, particularly regarding the rights to privacy and due process. It also aligns with National Law No. 25.326 on the Protection of Personal Data, ensuring that all AI-driven data processing within the courts adheres to federal standards. Furthermore, the protocol is intended to be read in conjunction with the existing 'Code of Civil and Commercial Procedure' and 'Code of Criminal Procedure' of the province, providing a specialized layer of rules for the digital aspects of these proceedings. The document also acknowledges the influence of international soft law and regional agreements. It references the principles established by the Inter-American Commission on Human Rights regarding technology and freedom of expression. By situating itself within this network of legal instruments, the protocol ensures that it does not create legal conflicts but rather fills a specific regulatory gap. It serves as a localized implementation of broader digital justice initiatives promoted by the 'Junta Federal de Cortes y Superiores Tribunales de Justicia de las Provincias Argentinas y Ciudad Autónoma de Buenos Aires' (JUFEJUS), contributing to a harmonized approach to judicial AI across the Argentine provinces. This alignment ensures that decisions made in San Juan are consistent with national legal trends and can withstand scrutiny at higher levels of the judiciary.

International Alignment

The protocol demonstrates a high degree of alignment with international best practices for AI governance. It specifically incorporates the 'Ethical Principles for the Use of Artificial Intelligence in Judicial Systems' developed by the European Commission for the Efficiency of Justice (CEPEJ). These principles include the respect for fundamental rights, non-discrimination, quality and security, transparency, and the principle of 'under user control.' By adopting these internationally recognized standards, San Juan ensures that its judicial AI framework is robust and compatible with global norms, which is essential for maintaining international legal cooperation and trust. Furthermore, the protocol reflects the recommendations of the OECD on Artificial Intelligence, particularly the emphasis on 'Trustworthy AI.' This involves a commitment to transparency and responsible disclosure regarding AI systems. The San Juan Court of Justice's focus on human-centric AI aligns with the G20 AI Principles, which Argentina has endorsed. This international alignment not only provides a solid theoretical foundation for the provincial regulation but also facilitates the exchange of knowledge and technology with other jurisdictions globally. It signals to the international community that the San Juan judiciary is committed to a modernization process that is both innovative and ethically grounded, positioning the province as a leader in the responsible use of AI in the public sector.

Implementation Timeline

Compliance Checklist

Sources and References