Buenos Aires Province AI Use Guidelines for Public Administration

Overview

Buenos Aires Province Resolution 4/2025, titled 'Directrices de uso de Inteligencia Artificial Generativa en la Administración Pública de la Provincia de Buenos Aires,' represents a foundational regulatory instrument designed to govern the deployment and utilization of Generative Artificial Intelligence (GenAI) within the provincial public sector. Enacted on February 10, 2025, and entering into force upon its publication on February 14, 2025, this resolution underscores the provincial government's proactive stance in addressing the transformative potential and inherent risks associated with advanced AI technologies. The primary objective of these guidelines is to establish a clear framework that ensures the responsible, ethical, and secure integration of GenAI tools into public administration functions. This initiative is driven by a commitment to safeguard fundamental human rights, protect personal data, and foster an environment of transparency and accountability in digital governance.

The resolution specifically targets all personnel and agents, regardless of their employment modality, who interact with GenAI systems in the course of their duties or services provided to the public administration of Buenos Aires Province. It emphasizes a preventative and demonstrably responsible approach, aiming to mitigate potential harms and ensure that technological innovation never compromises the rights and well-being of citizens. The guidelines are structured around four core principles: responsibility, human oversight, continuous training, and impact evaluation. These principles are intended to guide the selection, implementation, and ongoing use of GenAI, ensuring that these powerful tools serve the public interest while upholding democratic values and ethical standards. By setting forth minimum criteria and recommendations, the resolution seeks to cultivate a conscious and informed use of GenAI, thereby maximizing its benefits while effectively managing its associated challenges.

Definitions

While the resolution itself does not provide an exhaustive standalone glossary within the available snippets, it implicitly defines key terms through its context and application. 'Inteligencia Artificial Generativa' (IAGen or GenAI) refers to artificial intelligence systems capable of generating new content, such as text, images, audio, or video, based on learned patterns from existing data. This encompasses a broad range of AI models, including large language models (LLMs) and other generative adversarial networks (GANs) or transformers, which are increasingly being adopted for various administrative tasks. The guidelines are specifically tailored to the unique characteristics and potential implications of these generative capabilities, distinguishing them from more traditional, analytical AI systems.

The term 'agentes' or 'personal' (agents or personnel) is used to broadly define the scope of individuals covered by these directives. This includes all persons, irrespective of their employment or contractual status, who utilize or interact with GenAI tools within the framework of their official duties or services rendered to the public administration of the Province of Buenos Aires. This comprehensive definition ensures that the ethical and security obligations extend to every individual who might engage with GenAI on behalf of the provincial government, thereby establishing a consistent standard of conduct across the entire public sector workforce. The resolution also implicitly categorizes AI systems by risk levels, mentioning 'risk-limited' systems like generic chatbots (provided they are transparent and don't process sensitive data) and 'null-risk' systems like spam filters, indicating a nuanced approach to regulation based on potential impact on fundamental rights and safety.

Governance and Institutional Framework

The governance framework for the implementation and oversight of these GenAI guidelines is primarily vested in the Subsecretaría de Gobierno Digital (Undersecretariat of Digital Government) and the Dirección Provincial de Innovación Digital (Provincial Directorate of Digital Innovation) within the Ministry of Government of the Province of Buenos Aires. These bodies are explicitly tasked with the mission of implementing public policies related to the digital transformation of the province, including the crucial responsibility of elaborating ethical and regulatory aspects concerning artificial intelligence. The Dirección Provincial de Innovación Digital, in particular, was instrumental in publishing these guidelines, acting as the central authority for setting criteria and minimum standards for GenAI use across the public administration.

The resolution establishes a clear institutional process for the authorization and responsible use of GenAI tools. It mandates that the use of GenAI must be restricted to specific tasks within the relevant government agency. Furthermore, only authorized personnel are permitted to utilize these tools, requiring access via their institutional email and obtaining prior, formal authorization from their immediate hierarchical superior. This authorization must be formalized through a note in the GDEBA system, the provincial electronic document management system. This hierarchical approval mechanism ensures a structured approach to GenAI adoption, embedding accountability at various levels of the administration. The Subsecretaría de Gobierno Digital is also designated as the rector body in matters of telecommunications, digitalization, informatics, and technological solutions, reinforcing its central role in guiding the province's digital strategy, including AI integration.

Key Focus Areas

The Buenos Aires Generative AI Guidelines delineate several critical focus areas to ensure the responsible and ethical integration of GenAI into public administration. A paramount concern is ethical behavior and responsibility, requiring agents to utilize GenAI systems with principles of commitment, confidentiality, dignity, and respect for their professional functions. This ethical imperative forms the bedrock of all interactions with AI, ensuring that technology serves human values. Another crucial area is human oversight, which mandates human control both before and after the use of GenAI, explicitly prohibiting the delegation of sensitive decision-making entirely to AI systems. This ensures that human judgment and accountability remain central to public service delivery.

Furthermore, transparency and disclosure are emphasized, recommending the selection of AI tools that offer clear information about their functionality and scope, enabling users to comprehend their operations. The guidelines also stipulate that results generated by AI, particularly images, sound, video, or documents where AI is the creator, must carry a clear warning or label indicating their AI origin. Confidentiality and data protection are rigorously addressed, requiring all information handled by agents to be treated as confidential. Crucially, any confidential provincial data or personal/sensitive data identifying individuals must be anonymized before being shared with GenAI tools. Agents are held responsible for the access and handling of information and IT resources. In terms of information security, the resolution strictly prohibits the input of information that could compromise the security of the Province of Buenos Aires, such as passwords or critical infrastructure details, and mandates immediate reporting of any security incidents or inappropriate use. Finally, continuous training and literacy are promoted to ensure agents understand the limitations and potential biases of GenAI, fostering a culture of informed and critical engagement with these technologies.

Implementation Framework

The implementation framework for the GenAI guidelines in Buenos Aires Province is designed to integrate these directives into the daily operations of public administration agents through a structured and accountable process. A fundamental aspect is the authorization of access and use, where agents must secure explicit approval from their immediate hierarchical superior before utilizing any GenAI tool for official tasks. This approval is formalized through the GDEBA system, ensuring a verifiable record of authorization. Access to these tools is strictly limited to institutional email accounts, reinforcing security and accountability measures. This centralized control over access and authorization aims to prevent unauthorized or inappropriate use of GenAI, ensuring that its deployment aligns with official mandates and ethical considerations.

Beyond initial authorization, the framework emphasizes ongoing compliance and responsible practices. Agents are required to fully understand and adhere to the privacy and data protection policies associated with any digital instrument they use, particularly GenAI tools. A critical obligation is the anonymization of confidential or personal data before it is input into GenAI systems, safeguarding sensitive information and upholding data subject rights. The guidelines also stress the importance of selecting tools that offer options to limit data retention, further enhancing privacy. Furthermore, agents are mandated to report immediately any security incidents or instances of inappropriate use to the designated provincial authorities, such as the Dirección Provincial de Sistemas Informáticos y Tecnología. This reporting mechanism is vital for maintaining the integrity and security of the provincial digital infrastructure and for promptly addressing any deviations from the established guidelines.

Monitoring and Evaluation

The Resolution 4/2025 emphasizes the importance of robust monitoring and evaluation mechanisms to ensure the ongoing effectiveness and adherence to the Generative AI guidelines within the Buenos Aires public administration. This commitment reflects a dynamic approach to AI governance, recognizing that the landscape of AI technology is constantly evolving and requires continuous assessment. The guidelines stipulate that impact assessments should be conducted for GenAI systems, tailored to their specific application scope. These assessments are crucial for identifying potential risks, biases, and unintended consequences, allowing for timely adjustments and mitigation strategies. The focus on evaluation underscores the proactive stance of the province in managing the societal and ethical implications of AI.

The implementation of GenAI in the public sector is envisioned as a phased adoption process, which inherently includes stages for monitoring and evaluation. This typically begins with pilot programs involving specific departments or use cases, followed by an evaluation of their effectiveness and potential for broader extension. A subsequent phase involves the incorporation of new use cases and a wider rollout. Crucially, the final stage entails the establishment of a continuous monitoring system equipped with indicators of success and risk. This system is designed to measure the impact of GenAI on operational efficiency, the quality of work, and, most importantly, the protection of fundamental rights. This ongoing oversight ensures that the use of GenAI remains aligned with the principles of transparency, equity, and security, allowing the administration to adapt its strategies based on real-world performance and emerging challenges.

Penalties, Liability, and Appeals

While the detailed provisions regarding specific penalties and a comprehensive liability framework are not extensively detailed in the publicly available summaries of Resolution 4/2025, the document clearly establishes a foundation for accountability. It is explicitly stated that non-compliance with the established guidelines or the failure to subscribe to required declarations will be considered a disciplinary fault. This provision underscores the seriousness with which the provincial government views adherence to these GenAI directives, indicating that violations can lead to internal administrative sanctions against public agents. The inclusion of such a clause is crucial for ensuring enforcement and fostering a culture of responsibility among personnel utilizing AI tools.

The framework implies that liability for misuse or harm caused by GenAI applications would likely follow existing administrative law principles, where public officials are accountable for their actions and decisions within their scope of duties. While the resolution itself may not introduce a novel liability regime specifically for AI, it sets the behavioral and operational standards against which compliance and responsible conduct will be judged. Any appeals regarding disciplinary actions would presumably follow established administrative appeal processes within the provincial government structure. The emphasis on human oversight and responsibility further suggests that ultimate accountability rests with the human agents and their superiors, rather than being solely attributed to the AI system itself, aligning with common ethical AI principles globally. The requirement for impact assessments and risk mitigation also serves to preemptively address potential liabilities by ensuring due diligence in AI deployment.

Relationship to Other Instruments

Buenos Aires Province Resolution 4/2025 does not exist in a vacuum but is framed within a broader legal and policy landscape, both nationally and internationally. The guidelines explicitly acknowledge and are intended to be consistent with the Argentine National Constitution, particularly articles 14, 17, 18, 19, 42, and 43, which pertain to fundamental rights, property, privacy, consumer protection, and the right to information. This constitutional alignment ensures that the deployment of GenAI respects the foundational legal principles of the nation.

Furthermore, the resolution builds upon and complements existing national initiatives related to AI governance in Argentina. It references prior instruments such as Resolution N° 90/2021 (establishing the 'Programa de Inteligencia Artificial'), Disposition N° 2/2023 ('Recomendaciones para una Inteligencia Artificial Fiable'), and Resolution N° 161/2023 ('Programa de Transparencia y Protección de Datos Personales en el uso de la Inteligencia Artificial'). Additionally, Administrative Decisions N° 750/2023 and N° 899/2024, which created and regulate the Interministerial Board on Artificial Intelligence under the Chief of Cabinet of Ministers, provide a federal context for AI policy coordination. These references indicate a coordinated approach to AI regulation across different levels of government, ensuring coherence and mutual reinforcement among various policy instruments. The provincial guidelines thus serve as a specific implementation layer within this broader national strategy.

International Alignment

The Buenos Aires Province Resolution 4/2025 demonstrates a clear awareness of and alignment with international developments in AI governance, particularly those emanating from the European Union. The guidelines explicitly take into account and draw inspiration from the European Declaration on Digital Rights and Principles for the Digital Decade. This reference highlights a commitment to universal human rights and digital ethics as guiding principles for AI development and deployment. The Declaration, which emphasizes human-centric, ethical, and sustainable digital transformation, provides a foundational philosophical underpinning for the provincial directives.

Moreover, the resolution makes a direct reference to Regulation (EU) 2024/1689, commonly known as the EU AI Act, which establishes harmonized rules on artificial intelligence and is mandatory for EU member countries. This citation is significant as it indicates that Buenos Aires Province is looking to leading international regulatory frameworks, particularly those adopting a risk-based approach, to inform its own policies. By acknowledging and considering such comprehensive international regulations, the province signals its intent to ensure that its GenAI guidelines are not only locally relevant but also globally compatible with best practices for balancing innovation with ethical considerations and individual rights. This international alignment promotes cross-border cooperation and fosters a shared understanding of responsible AI governance.

Implementation Timeline

Sources and References