Democratic Algorithmic Accountability Framework (DAAF): A Methodological and Engineering Model for Algorithmic Governance and AI Act Compliance in the Social Economy

Abstract

Executive Summary Regulation (EU) 2024/1689, the Artificial Intelligence Act ("AI Act"), together with the GDPR and the Data Governance Act, imposes upon organisations a structural redesign of the management of algorithmic systems, particularly where such systems are deployed in workforce management and in high-impact decision-making processes. For the cooperative and social economy, this transition is not merely a matter of regulatory compliance. It constitutes a challenge bearing upon the very identity of the organisation: preserving the foundational democratic principle, "one member, one vote", and the centrality of the human person in the face of the opacity of machine learning models. The scholarly literature and the frameworks available on the market reveal an evident methodological gap. On the one hand, there exist rather abstract technical-organisational compliance standards, such as ISO/IEC 42001, or ethical mapping exercises devoid of genuine operational traction. On the other hand, integrated models capable of translating the legal requirements of human oversight into engineering workflows consistent with mutualistic principles are lacking. The Democratic Algorithmic Accountability Framework (DAAF) has been developed to bridge this divide by integrating three disciplines: public and administrative law, through the extension of the meta-principle of algorithmic legality; management engineering, through Business Process Reengineering and the Plan-Do-Check-Act (PDCA) cycle; and data governance, through immutable audit registers and explainability protocols designated as TRACE. The framework has been empirically validated through the operational platform AGORA, whose recommendation engine has accumulated over 56,000 immutable audit logs across a 19-month horizon.