Argentina - AI Regulation Framework (2505-D-2023)

Overview

Expediente 2505-D-2023, titled "Marco legal para la regulación del desarrollo y uso de la Inteligencia Artificial," is a bill introduced in the Argentine Chamber of Deputies on 8 June 2023 by Deputy Victoria Morales Gorleri. The bill seeks to create a national legal framework that governs research, development, deployment and use of AI across the Argentine territory. Its stated aims include protecting human rights, privacy and security while promoting ethical, transparent and responsible innovation. The bill is published in the official parliamentary docket and the original project text is available as a primary source in the Honorable Cámara de Diputados repository (Project text (PDF)) and summarized on the Chamber’s project page (Project page). The draft establishes general principles, a risk-based classification of AI uses, requirements for documentation, registration and certification and empowers the Executive to determine the authority of application. It also cross-references national scientific governance bodies such as the Gabinete Científico y Tecnológico (GACTEC) in relation to registration and verification duties (Law 25.467 / GACTEC references).

Definitions

The bill defines core terms to delimit its scope: "Artificial Intelligence (AI)" as computational systems designed to perform tasks requiring human-like abilities (learning, perception, reasoning, decision-making); "Autonomous AI System (SIAA)" as systems capable of operating without direct human intervention; "algorithm" as the set of instructions for problem solving and decision-making based on data and mathematical rules; "personal data" consistent with Argentina’s data-protection law; and "automated decisions" as decisions taken by algorithms or AI models without direct human involvement. These definitions are broadly aligned with contemporary regulatory drafting approaches and are used to shape obligations (for example, which systems require evaluation, documentation and registration).

Governance and Institutional Framework

The bill assigns the Poder Ejecutivo Nacional the task of designating the authority of application for the statute. It contemplates coordination with national science and technology governance mechanisms, specifically invoking registration and verification through bodies tied to the national science framework established under Law 25.467 and the Gabinete Científico y Tecnológico (GACTEC). Practically, that creates a two-layer governance model: a regulatory/administrative authority to implement, monitor and sanction plus technical-scientific actors (committees, testing bodies, or GACTEC-linked entities) responsible for certification, technical evaluation and registration. The text anticipates inter-institutional committees and ethical review boards to provide multidisciplinary oversight. This placement anticipates cooperation with existing administrative initiatives such as the Agencia de Acceso a la Información Pública’s "Program of transparency and protection of personal data in the use of AI" (Resolution 161/2023) (AAIP Resolution 161/2023), and provides a legal channel to align administrative programs and national plans for science and technology.

Key Focus Areas

The bill’s principal thematic pillars are: (1) Principles: it codifies ethical principles (respect for human dignity, privacy, transparency, responsibility, and equity) and an explicit ban on illegal, discriminatory or malicious uses of AI; (2) Risk Management: it mandates AI impact assessments and a four-tier risk classification (unacceptable, high, limited, insignificant) to determine prior obligations and prohibitions; (3) Transparency & Explainability: it requires documentation and disclosure of algorithmic functioning and grants affected individuals the right to request explanations of decisions that materially affect them; (4) Data Protection: it instructs compliance with existing data protection law (Law No. 25.326), including collection, lawful basis and informed consent requirements; (5) Security & Robustness: it requires technical and organizational safeguards to prevent unauthorized access, tampering or malicious interference; (6) Verification, Certification and Registration: it demands registration with a designated scientific/technical authority and subjecting systems to verification and certification processes; (7) Institutional Oversight and Testing: it promotes creation of committees and laboratories for ethical and technical testing; and (8) Sanctions & Remedies: it contemplates liability, penalties and administrative measures for non-compliance. The bill thus adopts a broad regulatory approach combining substantive prohibitions, procedural duties and ex-ante evaluations to manage AI-related harms.

Implementation Framework

Implementation under the bill is envisioned as a hybrid administrative-technical regime. The Executive designates the authority of application and issues implementing regulations to set out procedural rules, forms, timelines and technical requirements. Developers and providers must register AI systems (at particular development lifecycle stages) with the designated authority or the national scientific registry referenced in Law 25.467, provide technical documentation, evidence of impact assessments, risk mitigation measures and certification evidence. The bill requires periodic reassessment of risk classifications and updating of impact assessments. It also anticipates the creation or designation of accredited testbeds, ethics committees and conformity-assessment bodies to perform independent verification. The bill refers to the need for technical standards and interoperability where appropriate and leaves implementation details — such as fees, processing times and forms — to secondary regulation issued by the authority and coordinating scientific bodies.

Monitoring and Evaluation

Monitoring and evaluation are structured around registration, reporting and periodic audits. Registered systems are subject to ongoing oversight, requiring holders to update risk assessments, report incidents or adverse effects, and permit audits by the authority or accredited evaluators. The bill enables the production of state reports and public registers documenting certified systems and transparency data, which supports accountability and public oversight. It also contemplates metrics and periodic reviews of the law’s operation, providing for updating regulatory instruments to keep pace with technical change. The bill’s design anticipates coordination with administrative programs such as AAIP’s transparency program to monitor public-sector AI use and to publish transparency criteria.

Penalties, Liability, and Appeals

The bill articulates developer, provider and user responsibilities and contemplates administrative sanctions for breaches of registration, certification, transparency and data-protection duties. It provides for examination of errors of use and negligence, creates administrative enforcement powers for the designated authority and preserves civil liability remedies for damages caused by AI systems. Prohibited uses (e.g., systems that violate human rights or cause grave harm) may be banned and subject to sanctions. The measure contemplates procedural safeguards, including rights to administrative appeal and judicial review consistent with Argentine administrative law. Specific sanctions, monetary fines and procedural rules are expected to be specified in implementing regulations.

Relationship to Other Instruments

The bill cross-references and is designed to operate alongside existing Argentine instruments: Law No. 25.326 (personal data protection), Law No. 27.275 (access to public information), Law No. 25.467 (Science, Technology and Innovation framework and GACTEC), and administrative measures such as AAIP Resolution 161/2023 on transparency and data protection in AI use. It is explicitly positioned to complement national science policy instruments and administrative programs (for example, the AAIP Program and ministerial interministerial coordination platforms like the Mesa Interministerial sobre Inteligencia Artificial created by decisions of the Jefatura de Gabinete). The bill does not, on its face, supplant sectoral laws but creates cross-cutting duties applicable to AI in regulated sectors (healthcare, finance, public administration) while leaving room for sector-specific implementing rules and international commitments to inform implementation.

International Alignment

The bill references, and its principles are consistent with, international frameworks such as the OECD AI Principles and UNESCO's Recommendation on the Ethics of AI. The text and related administrative measures explicitly invoke those instruments as guiding references for transparency, explainability, privacy and human-rights protection. Argentina’s adoption of administrative programs (AAIP Resolution 161/2023) and the invocation of GACTEC situate the bill within an international alignment strategy intended to enable interoperability with regional and global norms while preserving local safeguards and human-rights protections.

Implementation Timeline

Sources and References