After the Zárate case, the Buenos Aires provincial government imposed strict regulations that require registering, auditing, and limiting the use of AI
Compartir:
The Province of Buenos Aires has approved a mandatory framework to regulate the use of Artificial Intelligence. The measure comes days after the Zárate case, where the municipality appointed a chatbot —“ZARA”— as General Director of Citizen Services.
Matzkin stated on his social network X: "We're not only betting on A.I. for industrial promotion, but we also trust this technology to improve every area of management and our city."
Marcelo Matzkin, intendente de Zárate, firmando el Decreto Municipal Nº 532/2025
This sparked a strong public debate and Kicillof replied with a resolution published in the Official Gazette that imposes strict controls throughout the provincial administration.
The regulation covers “the development, creation, research, innovation, contracting, acquisition, use and/or deployment” of AI. According to the Undersecretariat of Digital Government, all agencies must comply with mandatory requirements for security, transparency, fairness, sustainability, and human oversight.
This is not a guide or a recommendation. This is an auditable regime with penalties in case of non-compliance.
Risk framework
The resolution adopts a classification system inspired by the European Union's AI Act. It establishes four categories: unacceptable risk, high, limited, and none.
What's notable is that unacceptable risk systems are directly prohibited in the Province of Buenos Aires. These include those that may affect safety, health, essential rights, or the “constitutional-democratic order.”
The text also prohibits systems that perform “classification and/or social scoring”. This includes any mechanism of automated evaluation of citizen behavior to condition benefits, reputation, or access to public services.
El Municipio de Zárate creó un chatbot oficial denominado “ZARA” y lo designó como “funcionaria no humana”
In the high-risk category are biometric identification systems, those that manage critical infrastructure, and those used in education, justice, and public administration. All of them will require a prior algorithmic impact assessment.
Limited risk systems —such as chatbots or automated assistants— must explicitly inform citizens when they're interacting with a machine. No-risk systems, such as spam filters, must comply with good technical practices.
Obligations, registries, and ongoing oversight
The State must oversee the entire life cycle of each AI system. Public agencies will be required to conduct preliminary risk assessments, apply periodic management strategies, and "document any bias or discrimination."
They must also ensure the preservation of data “for as long as necessary” to allow audits.
When personal data is processed, it will be mandatory to have “free, express, prior, unequivocal, specific, and informed consent.” If there is interaction with citizens, the Government must warn from the outset that the response comes from an automated system.
Another key point is the mandatory registry. Each agency must register its AI systems in a provincial registry managed by the Undersecretariat of Digital Government.
Subsecretaria de Gobierno Digital en PBA, Sandra D'Agostino, a cargo de la regulación de la IA.
The regulation incorporates an unprecedented environmental criterion. It requires AI systems to be “sustainable and environmentally friendly” and mandates measures to prevent ecological damage. This includes energy consumption, carbon footprint, and the use of resources associated with model training.
A brake on development before it begins
The regulation entails a fierce intervention in a technology that could reduce costs, streamline procedures, and modernize public administration. However, the Buenos Aires government chose to impose strict controls and regulations before development takes off.
The province is creating a restrictive framework that limits innovation, increases bureaucracy, and places agencies under threat of sanctions.
At a time when AI is advancing worldwide, Kicillof's regulation appears as a brake that conditions its adoption in the public and private sectors. The tool that could improve state efficiency and open new productive opportunities is now subject to a framework that discourages its use from day one.
