Public-sector AI ethics: Italy’s governance debate meets France Travail’s algorithmic targeting

An Agenda Digitale essay argues that public-sector AI should be guided by human dignity, supervision, the common good, information quality and the centrality of work. That abstract framework becomes concrete in a separate report on France Travail, where an internal algorithm uses 26 variables and training data from 60,000 past controls to decide which unemployed people are prioritized for inspection. The story shows how quickly automated scoring can become a tool for profiling if transparency and accountability are weak. For Italian public administrations and the vendors that serve them, the lesson is that operational ethics is not a slogan: it must be built into decision flows, documentation and appeal mechanisms. It is also relevant for DPOs and legal teams because these systems can materially affect rights and access to services. The broader implication is that public-sector AI needs governance before scale.

Sources

• Dalla Magnifica Humanitas alla PA: un’etica operativa per l’AI — Agenda Digitale
• France Travail – 26 variables décident si vous êtes suspect — Korben