Data and AI Governance as Tools to Manage Risks Arising from Bias
Keywords:
Governance, AI Governance, Bias in AI, Risk Management and AI, ISO 42001Abstract
This article addresses the concept of bias, distinguishing it from other related concepts with which it is sometimes confused. It describes the stages of the AI systems lifecycle, identifying how bias can penetrate each stage in order to manage the resulting risks.
After an overview of the obligations imposed on different stakeholders by the RIA and which can enable the management of risks to fundamental rights, we have focused on two specific aspects. First, data governance, which refers to how the data used by high-risk AI systems is managed, and which has a direct impact on bias. Second, AI governance allows the various obligations to be organized around a management system in such a way that they are planned, operated, monitored and continuously improved, for which it is recommended to use a standard such as BS ISO/IEC 42001:2023 (Information technology — Artificial intelligence — Management system, now a Spanish standard through UNE-ISO/IEC: 2025), due to the benefits mentioned in the article.
