News | AI with Rights | Digital Platforms and Markets

AI regulation must take into account national context, argues Data Privacy

For the Data Privacy Brazil Association, any regulation of Artificial Intelligence (AI) in the country must take into account Brazilian socioeconomic conditions, while seeking convergence with international regulatory models. In this sense, the entity considers Bill 2338/2023 to be the best proposal for this purpose currently being processed in the National Congress, although it could be improved, especially with regard to anti-discrimination criteria and training the population to use the technology.

The evaluations are part of the study titled “Central Themes in AI Regulation: Local, Regional, and Global in the Pursuit of Regulatory Interoperabilit”, recently published by the association.

In general terms, the NGO dedicated to data protection emphasizes that “even with the possibilities of refinements in Bill 2338, the project is, at least today, a less bumpy path for establishing AI governance in line with the Brazilian socio-economic context”.

The organization also states that the text “represents a first step” toward regulating AI “from a human-centered approach”, taking into account the experience of living in Brazil with its “structural asymmetries and inequalities”, including racism.

Suggestions

In the association’s assessment, the Bill has the virtue of acknowledging the inequalities and structural asymmetries of the brazilian context. However, it points out that, within this theme, the text could go further in its anti-racist and anti-discriminatory commitment. One example would be to include the “explicit ban on AI systems considered racist, sexist, and transphobic, especially in sensitive contexts”, such as facial recognition applications for public security.

Additionally, Data Privacy suggests that the proposal could include a chapter on the duties of the Public Power, aimed at encouraging “investment in the training of the population for the safe, conscious, and responsible use and development of AI systems.”

For the association, the text still adopts a defensive stance on governance regarding protection against illegitimate or illegal outcomes produced by AI models. Therefore, it suggests that the Bill advance in reactive proposals, such as promoting the development of ethical, diverse, open, and multidisciplinary AI databases and systems within the country.

Nonetheless, the organization emphasizes that national regulation cannot simply rely on regulatory models developed abroad, ignoring the country’s particularities.

“It is crucial that Brazil develops an AI regulation that shares similarities with foreign discussions and models to ensure regulatory convergence, but taking into account the country’s specificities so that the technology regulation works for the Brazilian context and the people living here, as initiated by Bill 2338/2023,” the association points out.

The text was originally published on December 12, 2023, on the Tele.síntese.