The year 2023 was, in a way, the regulatory year for Artificial Intelligence (AI). Still in May, the G7 Summit highlighted the importance of promoting guardrails for advanced AI systems on a global basis.
In August, it was China's turn to enact a law specifically associated with Generative AI, with the aim of mitigating essential harm to individuals, maintaining social stability and ensuring its long-term international regulatory leadership.
In the wake of this process, it was up to the US, represented by its then president Biden, to issue an executive order that was responsible for guiding the application of AI in the field of reliability, security and the protection of the fundamental elements of American sovereignty.
However, the cherry on top was largely the EU AI Act, pre-approved in December 2023 and enacted in early 2024. Deeply debated and quite comprehensive, the Act attains the status of a regulation with an international vocation conceived as a legal framework for the development and application of AI systems for the member countries of the bloc.
In Brazil, Law 2,338 on Artificial Intelligence marks a turning point in the regulation of emerging technologies in the country. On a large scale, the law has positive aspects, but it also shows some fragility in strategic areas for the development of our leadership in the field of AI.
At the center of Brazilian regulation are provisions of the General Data Protection Law (LGPD), emphasizing the protection of personal data with a focus on privacy. The law aims, thus, to ensure that AI does not compromise individual rights. The LGPD also aims to encourage innovation by offering some tax incentives and subsidies to companies that invest in AI research and development. This aspect aims to position Brazil as a technological innovation hub, stimulating competitiveness and the creation of startups in the AI sector. Regarding social impacts, digital inclusion and the ethical use of AI to reduce inequalities are addressed through the promotion of educational and training programs for vulnerable populations, preparing the workforce for the artificial intelligence era. The idea is to mitigate the negative social impacts of automation, promoting a more equitable transition.
However, there are negative points to highlight. The first of them revolves around excessive bureaucracy, such as the requirement for multiple assessments and certifications that could burden companies—especially startups and small businesses—with additional costs and lengthy processes. This bureaucratic aspect can discourage innovation and the adoption of new technologies. Although the law has interesting intentions, some critics cite ambiguity in certain provisions, allowing for conflicting interpretations and legal uncertainty. There is a lack of clarity regarding responsibilities and specific penalties that will hinder their practical application. There are still concerns about the potential use of AI regulation for government control purposes. This aspect raises questions about the protection of civil liberties and the limits of state intervention.
We are, in any case, facing an important milestone in AI regulation. Such a regulatory component is necessary to bring balance between rights protection, innovation encouragement, and social inclusion promotion. However, the effectiveness of the law will depend on its practical implementation and the ability to mitigate the associated risks. Transparency, regulatory clarity, and constant civil society oversight will be essential to ensure that benefits outweigh challenges.
