Argentina: Proposed legal framework for the regulation of the development and use of Artificial Intelligence [original language] - Raw

Preamble

DRAFT LAW The Senate and Chamber of Deputies of the Argentine Nation, gathered in Congress, enact with the force of Law.

Legal framework for the regulation of the development and use of Artificial Intelligence

TITLE I General Provisions

Article 1: Object of the law

To establish the legal framework for the research, development, use and regulation of artificial intelligence in the national territory, in order to guarantee the protection of human rights, privacy and security of citizens, promoting transparency, ethics and responsibility in its application.

To foster innovation and progress in the field of AI, promoting human benefit and avoiding risks and negative consequences.

Article 2: Authority and Scope of application.

2.1. The National Executive Power shall determine the enforcement authority of this law.

This law shall apply to all natural or legal persons who develop, research, innovate or use artificial intelligence in the Argentine territory, regardless of where the server or servers that support it are located.

Article 3: Definitions:

it shall be understood as:

3.1. Artificial Intelligence (AI): It refers to computational systems designed to perform tasks that require human skills, such as learning, perception, reasoning and decision making.

3.2. Autonomous Artificial Intelligence System (AAIS): It refers to an AI system capable of operating independently without direct human intervention.

3.3. Algorithm: set of instructions for solving a problem, performing a task or making decisions, based on data and mathematical rules.

3.4. Personal data: any information relating to an identified or identifiable natural person.

3.5. Automated decisions: they refer to choices or conclusions that are made by artificial intelligence systems or by algorithms without the direct intervention of a human being. These decisions are based on rules, algorithms or machine learning models that analyze and process data to determine the most appropriate action or result in a given situation.

3.6. Usage Error: any action, omission or negligence by the developers, suppliers or users of artificial intelligence systems that results in adverse consequences, damage or harm to persons, property or rights shall be considered a usage error of artificial intelligence.

TITLE II Principles

Article 4: Ethical Principles

4.1. All development and use of AI must be based on fundamental ethical principles, including respect for human dignity, privacy, transparency, responsibility and fairness.

4.2. The use of AI for illegal, discriminatory, malicious purposes or those that violate human rights is prohibited.

Article 5: Responsibility and Accountability

Developers, suppliers and users of artificial intelligence systems are responsible for the consequences of their actions and decisions related to the use of AI and must ensure that their systems are safe, reliable and comply with established quality standards.

5.2. The obligation to document and disclose the operation and algorithms used in AI systems is established, to allow auditing and evaluation of their impact.

5.3. Guarantee the quality and effectiveness of the services offered with artificial intelligence AND Assume responsibility for the decisions and actions of artificial intelligence systems.

Article 6: Privacy and Data Protection

6.1. AI systems must respect and protect user privacy and the processing of their personal data in accordance with applicable data protection regulations.

6.2. Within the framework of law 25326 on personal data protection, the unauthorized use of personal data collected by AI systems is prohibited, and the obligation to obtain the informed consent of individuals for the use of their data is established.

Article 7: Transparency and Explainability

7.1. AI systems are required to be transparent in their operation, so that users understand how decisions are made and results are reached.

7.2. The right of individuals to request explanations of decisions made by AI systems that affect them is established.

Article 8: Security and Robustness

8.1. AI systems must be designed and developed with adequate security measures to prevent unauthorized access, manipulation or malicious interference.

8.2. The implementation of technical and organizational safeguards to ensure the robustness and reliability of AI systems is required.

TITLE III Impact Assessment and Risk Management

Article 9: Impact assessment

9.1. AI impact assessments are required to identify and mitigate potential risks prior to deployment.

9.2. The creation of committees and bodies responsible for conducting ethical testing of AI systems will be promoted, in order to ensure their compliance with established principles.

Article 10: Risk assessment and management.

Those responsible for artificial intelligence systems shall be required to implement adequate measures to mitigate the risks identified during the risk assessment. Risk management shall include the adoption of transparency, accountability and security measures to minimize the impact of identified risks on fundamental rights and the safety of persons.

The assessment shall be updated periodically, and shall include a classification of the risks associated with the artificial intelligence system, according to the following risk levels:

a. Unacceptable risks: those risks that are considered unacceptable due to their severity or impact on fundamental rights or the safety of persons. The presence of an unacceptable risk in the artificial intelligence system will require the adoption of adequate mitigation measures or the prohibition of the system.

b. High risks: those risks that have a significant impact on the fundamental rights or safety of persons. The presence of a high risk in the artificial intelligence system will require the adoption of adequate mitigation measures.

c. Limited risks: those risks that have a minor impact on the fundamental rights or safety of persons, but still require mitigation measures.

d. Insignificant risks: those risks that have a minimal impact on the fundamental rights or safety of persons.

Article 11. Verification and certification.

Developers of AI systems, at any stage of their life cycle, must register them in accordance with the procedures established by GACTEC (Scientific and Technological Cabinet) and within the framework of Law 25.467 on Science, Technology and Innovation.

Artificial intelligence systems must undergo a verification and certification process to ensure their quality and safety. Those responsible must prove that the systems comply with the applicable technical and legal requirements, as well as with the ethical principles and obligations established in this law.

TITLE IV Prohibitions and/or restrictions

Article 12: Consider the prohibition and/or restriction of the use of AI systems in cases of:

Violation of human rights: when the artificial intelligence system is used to violate fundamental human rights, such as privacy, freedom of expression, equality or human dignity.

Unfair discrimination: when the artificial intelligence system is discriminatory and produces unfair or harmful results for certain groups of people, such as racial or gender discrimination.

Serious harm or safety risks: when it has the potential to cause serious harm to people, society or the environment.

Manipulation or undue influence: when it is used to manipulate or unduly influence important decision-making, such as in electoral processes or in the formation of public opinions, undermining the integrity and legitimacy of such processes.

Lack of transparency and accountability: when an AI system operates opaquely and it is not possible to audit or understand how it makes its decisions, making it difficult to assess its impact and hold the actors involved accountable. In such cases, a prohibition may be necessary until adequate transparency and accountability mechanisms are established.

In all cases where its use generates the perpetuation of inequalities or injustices.

The use of artificial intelligence systems in emergency situations must be justified and limited in time.

TITLE V Oversight and Control Authority

Article 13. Supervisory Authority

An Artificial Intelligence Supervisory Authority (ASIA) shall be established to ensure compliance with the law and the regulation of artificial intelligence systems. ASIA shall be an independent body and shall be composed of experts in artificial intelligence, ethics, law and other relevant fields and shall report to the authority determined by the National Executive Branch.

Article 14. Inspections and sanctions.

The ASIA shall:

a. The ASIA shall have the authority to conduct inspections, investigations and audits of artificial intelligence systems used in various sectors, including government, industry and public services;

b. maintain the public registry of intelligence systems with the procedures and requirements determined by the enforcement authority

c. receive and examine complaints and claims;

d. prohibit or restrict the use of artificial intelligence systems that violate the provisions established in this law or that present a significant risk to the human, civil or political rights of individuals;

e. require providers and developers of artificial intelligence systems to adopt specific measures to prevent and address discrimination and biases in the systems, including suspension or withdrawal of the system in case of non-compliance;

f. establish sanctions for non-compliance with the obligations established in this law, which may also include fines and removal from the registry of artificial intelligence systems, as well as temporary or permanent prohibitions on the use of artificial intelligence systems, and other corrective actions that are appropriate according to the severity and recurrence of the violation.

Article 15: International collaboration.

The national artificial intelligence supervisory authority shall coordinate with other international authorities in the field, in order to promote coordination and harmonization of policies and regulations in this area.

Article 16: Registration of AI systems

Those responsible for artificial intelligence systems shall be obliged to register their systems in a public registry created by the supervisory authority. The registry shall contain:

The technical characteristics of the artificial intelligence system, including its purpose and objectives, its design and operation, and the techniques and algorithms used.

The personal data they process and their origin, including the nature of the data, the sources of the data, and the recipients of the data.

The transparency, accountability and security measures applied to the artificial intelligence system, including the existence of a governance model for the system, the explainability and auditability of the system, and the technical and organizational security measures implemented.

The supervisory authority shall establish the procedures and requirements for the registration of artificial intelligence systems, and may establish categories of systems subject to different registration requirements, depending on their level of risk, taking into account criteria such as the purpose and objectives of the system, the nature of the data it processes and the degree of autonomy of the system.

Those responsible for artificial intelligence systems must update the information in the registry when significant changes occur in the systems or in their technical characteristics or in the data they process. The supervisory authority shall have access to the information in the registry to carry out its supervisory functions.

Developers and providers of artificial intelligence systems shall fully cooperate with ASIA and provide access to their systems, source codes, data and relevant documentation for the purposes of supervision and control.

TITLE VI International Cooperation and Standards

Article 17: International Cooperation

17.1. International cooperation shall be promoted for the development of common standards and regulations in the field of AI.

17.2. Collaboration between governments, international organizations, industry and civil society shall be encouraged to address challenges related to AI.

Article 18: Compliance and Sanctions

18.1. The competent authorities shall be responsible for enforcing this law and applying sanctions in case of non-compliance.

18.2. Sanctions for violation of this law may include fines, prohibition of use of AI systems, suspension of licenses and other appropriate measures.

CHAPTER VII Liability for damage or misuse

Article 19: Liability of Developers and Providers

19.1. Developers and providers of artificial intelligence systems shall be liable for misuse of their systems that cause damage. They shall be considered liable if they have not taken reasonable steps to prevent the error or if they have failed to comply with established regulations and standards.

19.2. Developers and providers shall implement adequate security measures and testing to minimize usage errors and ensure the quality and reliability of their systems.

19.3. In the event that a usage error occurs and the responsibility of the developers or providers is proven, they shall repair the damage caused, compensate the affected persons and take the necessary measures to prevent the error from recurring in the future.

Article 20: User Responsibility

20.1. Users of artificial intelligence systems shall be responsible for the proper use of such systems and for minimizing the associated risks. They shall use the systems in accordance with the instructions provided and in compliance with established regulations.

20.2. Users shall be diligent in the selection and configuration of artificial intelligence systems, considering their characteristics and limitations, and ensuring that they are used in an ethical and responsible manner.

20.3. In the event of a usage error by users that results in harm to third parties, users shall be liable for such harm and shall take the necessary steps to mitigate adverse consequences and compensate affected persons.

Article 21: Civil Liability Insurance

21.1. Developers, suppliers and users of artificial intelligence systems must have adequate civil liability insurance to cover possible damages caused by usage errors of the systems.

21.2. Civil liability insurance shall be provided by authorized insurance companies and shall cover both personal injury and property damage.

Article 22: Information and Education on Responsible Use

22.1. Developers, suppliers and users of artificial intelligence systems shall provide clear and understandable information on the responsible use of the systems, including limitations, risks and necessary precautions.

22.2. Education and training in the responsible use of artificial intelligence shall be promoted, raising awareness of the associated risks and best practices to avoid errors and minimize potential harm.

Article 23: Reporting of Usage Errors

23.1. A mechanism for reporting usage errors of artificial intelligence shall be established, where affected persons or any person who becomes aware of a usage error may report it to the competent authority.

23.2. Reports of usage errors shall be treated confidentially and the necessary measures shall be taken to investigate and correct the reported errors, as well as to apply the corresponding sanctions in case of negligence or non-compliance.

TITLE VIII Research, development and training

Article 24: Promotion of Research in Artificial Intelligence

24.1. Research in artificial intelligence (AI) shall be promoted with the objective of fostering scientific, technological and social advances in this field. Resources and incentives shall be allocated and a fund may be created, with public-private contributions, to support cutting-edge research, innovation and collaboration in AI.

24.2. Funding programs for AI research projects shall be established, both in academia and industry, with special emphasis on areas of strategic interest and public benefit.

24.3. The State may establish fiscal, financial and other incentives for research and development of artificial intelligence, especially for projects aimed at improving the well-being of society and promoting innovation.

24.4. AI research shall be conducted in accordance with ethical and legal principles, and transparency and knowledge sharing and results shall be encouraged to promote responsible and beneficial development for society.

Article 25: Responsible Development of Artificial Intelligence

25.1. The responsible development of AI shall be promoted, ensuring consideration of ethical, legal, social and safety aspects at all stages of the development of AI systems

25.2. AI system developers shall implement ethical design and social responsibility practices, considering fairness, transparency, privacy and security in the development of the systems.

25.3. Collaboration and dialogue between developers, researchers, users and society in general shall be encouraged to ensure that AI systems are developed in an inclusive manner and taking into account diverse perspectives.

Article 26: Training and Education in Artificial Intelligence

26.1. Training and education in AI shall be promoted for professionals, researchers and students, in order to develop skills and competencies necessary to understand, use and develop AI systems in an ethical and responsible manner.

26.2. AI education and training programs shall be established, both at the academic level and in the workplace, addressing technical, ethical, legal and social aspects of AI.

26.3. Collaboration between educational institutions, research centers and industry shall be encouraged to develop up-to-date and relevant training programs, adapted to the needs and advances in the field of AI.

Article 27: Knowledge Sharing and International Cooperation

27.1. The exchange of knowledge, data and best practices in the field of AI shall be encouraged at the national and international level, promoting cooperation between different countries, organizations and sectors.

27.2. International cooperation and collaboration agreements on AI shall be established, with the objective of promoting common standards, information exchange and joint development of policies and regulations.

27.3. Active participation in international initiatives addressing the ethical, legal and social challenges of AI shall be encouraged, seeking to influence the global agenda to ensure the responsible and sustainable development of AI worldwide.

TITLE IX Final Provisions

Article 28: Validity and Application

28.1. This law shall come into force upon its official publication and shall be mandatory throughout the national territory.

28.2. Artificial intelligence systems existing at the time of the entry into force of this law must comply with its provisions within a period established by the competent authority.

Article 29: Regulatory Adaptation

29.1. The competent authorities shall make the necessary modifications to existing regulations to ensure consistency and complementarity with this law regulating the development and use of artificial intelligence.

29.2. The competent authorities shall promote the dissemination and awareness of this law among relevant stakeholders, as well as provide guidance and technical support for its effective implementation.

Article 30: Evaluation and Update

30.1. The competent authority shall conduct periodic evaluations of the impact and effectiveness of this law, in order to identify possible improvements and necessary adjustments.

30.2. Based on the results of the evaluations mentioned in article 30.1, relevant updates and modifications shall be made to ensure the adequacy and effectiveness of the regulation of artificial intelligence.

Article 31: Protection of Fundamental Rights

31.1. The application of this law shall not compromise the fundamental rights and freedoms of individuals, including, but not limited to, privacy, equality and non-discrimination.

31.2. In case of conflicts between fundamental rights and the provisions of this law, an adequate balance and proportionality shall be applied, ensuring that rights are protected to the greatest extent possible.

Article 32: Repeal of Contrary Regulations

32.1. All regulations or provisions that are contrary to the provisions of this law regulating artificial intelligence are repealed.

32.2. The provisions of this law shall prevail in case of conflicts or contradictions with any other existing regulations, unless expressly stated otherwise.

Article 33: Budgetary provisions.

The National Executive Power is authorized to make the modifications and incorporations in the Law of budget of expenses and resources of the national administration for the current fiscal year in the aspects that are considered necessary for the implementation of this law.

Likewise, the incorporation of the necessary resources for the correct fulfillment of the functions of the enforcement authorities must be foreseen in the budget of the immediately following year.

Article 34: Communication.

Be it communicated to the National Executive Branch.