Regulation of Artificial Intelligence in the European Union ethical framework, risk classification and possible consequences in medicine

Abstract

Contextualization: The European Union's regulation of artificial intelligence seeks to establish a clear framework for classifying risks associated with these systems, covering categories of minimal, limited, high, and unacceptable risk. Problem: The central issue lies in ensuring that these systems do not compromise the safety and fundamental rights of individuals. On the other hand, excessive regulation, in addition to hindering technological development in fields such as medicine in some parts of the West, may become ineffective in the face of the competitiveness of major world powers. Objectives: This study aims to describe the risk classification methodology and the ethical principles guiding the implementation of a reliable artificial intelligence model. Methods: A documentary narrative review of the European regulation was used, with qualitative analysis to identify and categorize the specific requirements for each level of Artificial Intelligence risk, comparing the advantages and disadvantages of the technological revolution in medicine. Results: Artificial Intelligence systems classified as minimal risk, such as spam filters, are considered safe and do not require close supervision. High-risk systems, such as Artificial Intelligence for medical diagnosis, require strict compliance with ethical and safety standards, due to the potential impact on human life. In any case, given the potential technological advances in medicine and people's health, the ethical question arises of assessing how much legal regulation can delay new discoveries for individual and collective health. Final considerations: The final considerations indicate that the regulatory proposal significantly contributes to building a security model, but as long as this regulation is not global, it may be ineffective or even hinder scientific development in areas such as medicine.