Signal Applied and Filtered: An End-to-End Algorithmic Fairness Audit of A Public Employment Agency

Summary

Algorithmic fairness evaluation has commonly treated AI systems as technical components isolated from the organizational context in which they operate. To move beyond that limitation, this paper presents what the authors call the first independent end-to-end fairness audit, focused on a semi-automated hiring system run by Barcelona Activa, a public employment agency in Barcelona, Spain, which uses the third-party platform TalentClue for candidate search and shortlisting. Rather than examining a single algorithm in isolation, the study audits the entire hiring pipeline. Its aim is to surface fairness problems arising not only from the tool's internal logic but from how the public agency adopts and operates a third-party AI hiring tool. The findings offer implications for the procurement and oversight of algorithmic hiring systems in the European public sector.