Beyond the Promise of Innovation: The Hidden Risks of Artificial Intelligence in Healthcare

Abstract

Artificial intelligence has been transforming medicine by processing large volumes of clinical data through machine learning, optimizing diagnosis and care management within the hospital environment. However, the accelerated incorporation of these technologies at a pace faster than regulatory debate exposes the healthcare sector to serious ethical dilemmas, such as the lack of algorithmic transparency, algorithmic paternalism, loss of patient autonomy, clinical confirmation bias, and inequalities in healthcare access. Given this complex context, the overall objective of this study is to analyze the hidden, ethical, and operational risks associated with integrating artificial intelligence into healthcare delivery and management. To this end, a narrative review of a qualitative nature was conducted, focusing on the description and critical interpretation of scientific literature on the topic. The literature search was carried out in databases such as PubMed, SciELO, and Google Scholar, selecting works published between 2022 and 2026. The data indicate that over-reliance on automated systems compromises the human and empathetic dimension of care, while also exposing the healthcare system to legal vulnerabilities regarding civil liability in diagnostic failures and confidential data breaches. The unregulated proliferation of virtual assistants also encourages self-medication and distances patients from in-person care. Furthermore, the underrepresentation of vulnerable social groups in training datasets can reproduce biases and cause diagnostic errors that deepen disparities in care. It is concluded that technological advancement must be accompanied by the formulation of transparent governance guidelines, rigorous technical evaluation, and ongoing training for multidisciplinary teams, ensuring that human judgment remains the central pillar in medical decision-making to safeguard security, ethics, and quality of care.