Institutional governance of legal-substantive plugin vetting

Determine which authorities should endorse a plugin’s mapping to specific legal requirements, under what mandate and with what resources, and in coordination with which European-level infrastructure initiatives.

Background

The AI Assessment Sandbox Configurator uses a tiered Catalogue model in which plugins may eventually be placed in Core, Verified, or Community tiers. The Verified tier is intended to provide a quality signal for plugins that satisfy technical and documentation requirements. The authors raise the unresolved issue of whether verification should also include legal-substantive recognition of a plugin’s claimed mapping to particular provisions of the EU Artificial Intelligence Act or its Annexes.

Resolving this issue requires identifying competent endorsing authorities, defining their legal mandate and available resources, and determining how their role would interact with European-level infrastructure initiatives. The problem matters because downstream users, including Competent Authorities, would need a credible institutional basis for relying on a plugin’s claims about the legal requirements it addresses.

References

Which authorities should play this role, under what mandate, with what resources, and in coordination with which European-level infrastructure initiatives are open questions that will require broader discussion among \acp{CA}, EU-funded initiatives, and the assessment-tool community itself.

— The AI Assessment Sandbox Configurator: A Framework to Support Technical Assessment in AI Regulatory Sandboxes  (2610.01539 - Buscemi et al., 1 Oct 2026) in Section "Catalogue Governance", subsection "Catalogue Governance" (Section 6.2)