Escrever uma avaliação PREreview

Identifying the ultimate human actor responsible for the harm caused by AI systems remains one of the most urgent and unresolved challenges in AI governance. While existing literature emphasizes transparency, bias mitigation, and explainability, it often neglects the question of who is ultimately accountable for AI-enabled decisions and their consequences. This article introduces the concept of the Ultimate AI Accountability Owner (UAAO), a governance mechanism designed to close the accountability gap. The UAAO framework provides a structured approach for assigning final responsibility throughout the AI lifecycle, encompassing design, deployment, operation, and liability. Drawing on theories of accountability and risk governance, the paper presents a conceptual model supported by comparative case studies in hiring, finance, and healthcare. It argues that embedding UAAO roles within institutional governance enhances ethical oversight, clarifies accountability lines, and enables traceability in the event of failures. By addressing the persistent ‘responsibility vacuum,’ the UAAO framework offers a scalable solution for high-stakes AI deployment—ensuring that accountability remains human and institutionally embedded.