Guidelines for Using Generative Artificial Intelligence

The Provincial Directorate of Digital Innovation published a guide outlining minimum criteria and guidelines to be considered by all individuals and/or agents, regardless of their work or contractual modality, who use generative artificial intelligence (Gen AI) systems in the course of their functions or services provided to the public administration of the Province of Buenos Aires.

These recommendations aim to mitigate risks and encourage the conscious use of these technologies, with a focus on protecting human rights. To achieve this, the guide is structured around the following four principles: accountability, human oversight, continuous training, and impact assessment.

In this regard, the Directorate has established that the following criteria must be considered when using GenAI systems.

1. Authorization of access and use: The use of GenAI systems must be limited to specific tasks of the relevant agency. Only authorized personnel may use them, accessing the systems through their official institutional email and with prior approval, formalized by a note that their immediate supervisor submits through the GDEBA[1] system.

 

2. Ethical behavior: the personnel’s use of GenAI systems must be guided by principles such as accountability, commitment, confidentiality, and dignity.

 

3. Transparency: the recommendation is to choose tools that provide clear information about their functionality and scope, so that the personnel can understand how they operate.

 

4. Confidentiality and data protection: All information handled by the personnel when performing their duties must be considered confidential. Therefore, before entering data into a GenAI system, they must review its privacy policies and choose settings that limit data retention. It is also important to avoid including unnecessary information and to anonymize any sensitive, personal, or confidential data beforehand.

 

5. Information security: The personnel is strictly prohibited from entering into GenAI systems data that could compromise the security of the government of the Province of Buenos Aires. If any information security incidents are detected, they must be reported immediately to the appropriate authorities.

 

6. Evaluation of results: All content generated or modified using GenAI systems must be reviewed by qualified personnel before giving it any official use. This content should be considered a draft until it has been validated and, in cases involving the generation of images, audio, videos, or documents, its origin must be clearly indicated with a disclaimer.

 

7. Prohibited uses: GenAI systems cannot be used for illegal or offensive activities, for personal purposes, for data manipulation, unauthorized access, or any actions that could harm the systems or services of the government of the Province of Buenos Aires.

 

8. Awareness: the personnel is encouraged to stay informed about the limitations, best practices, and potential biases of GenAI systems, to promote more conscious and responsible use.