Another highlight is the risk-based regulatory approach , which proposes a risk assessment methodology adapted to different contexts and sectors. This means that regulation will not be the same for all types of AI, but will be calibrated according to the level of risk involved in each application, considering potential ethical, social and economic impacts.
Points of attention and challenges
Despite its merits, Bill 2338/2023 still has some weaknesses. The definition of limits for the use of high-risk technologies, such as facial recognition and predictive policing, is one of them.
The text suggests banning technologies such as autonomous belgium mobile database weapons and emotion recognition systems, but the lack of consensus on these issues indicates that a more in-depth debate is needed so that the legislation is adapted to the Brazilian reality. These systems, if poorly regulated, can increase inequalities and generate serious risks to citizens' privacy and security.
Another sensitive point is the creation of a competent authority to regulate AI, with sufficient powers and autonomy to monitor and enforce compliance with the legislation. The absence of a strong and well-defined regulatory framework could compromise the effectiveness of the law and make it difficult to apply sanctions in cases of inappropriate use.
Without these elements, regulation runs the risk of being merely declaratory, without a practical impact that guarantees security and the protection of citizens' rights.
Senate proposes new bill to regulate AI in Brazil. Credits: Jefferson Rudy - Senate Agency
For and against regulation: the innovation dilemma
The debate on AI regulation in Brazil is marked by a central dilemma: the balance between protecting fundamental rights and stimulating technological innovation. On the one hand, advocates of regulation argue that it is essential to ensure that the development and use of AI do not compromise the privacy, security and freedom of citizens.