Data Spaces and AI: Trustworthy Agentic Participation in Data Spaces

Two developments are reshaping how organizations work with data, and they are usually discussed separately. Data spaces give independent organizations a way to share data and data-based services under agreed rules, with each participant keeping control of their own assets. Artificial intelligence is being adopted across every sector, but its value depends on access to high-quality data with clear usage rights, reliable provenance and a sound basis for compliance. This paper connects the two, because each supplies what the other needs. In the AI era, the fundamental unit of exchange is progressively shifting from data to governed intelligence.

The relationship runs in both directions. Data spaces give AI a governed, high-quality data foundation. Catalogues make data discoverable across organizational boundaries, federated identity mechanisms extend trust where it cannot otherwise be assumed, and machine-readable policies make usage rights explicit and enforceable. In return, AI gives data spaces the automation they need to operate and scale, by generating metadata, aligning vocabularies, interpreting policies and reducing the manual effort that has kept cross-organizational data sharing slow and costly.

The frontier of this relationship is agentic participation. An AI agent can act through the same interfaces a human participant uses, faster and without interruption: finding data, negotiating terms and invoking services on an organization’s behalf. The central argument of this paper is that this can be made trustworthy without inventing new and untested machinery. An agent operates under a delegated identity that binds it to the legally accountable participant it acts for, so every action it takes remains attributable to a real organization and is governed by the same policies, credentials and audit trails that already secure human participation.

This is a position paper, not a legal opinion or a full technical specification. It establishes a shared vocabulary for readers coming from either side of this convergence, sets out today’s AI challenges and how the building blocks of a data space address them, details the concrete value AI brings to operating a data space and grounds the discussion in pilots already underway. For policy makers, it shows why data spaces belong within the agenda for AI governance and innovation. For practitioners on either side, it shows where AI creates new requirements and where it removes effort. The technical and governance foundations largely exist and are being standardized; the work now is to bring them together.