Money. Identity. Technology.
Follow the money. Find the decision.
Research inputs & evaluation
GDI contributed existing keyword resources; FBK edited D2.1 and recorded internal coordinator review. The document’s sampling and methodological choices are inspectable.
Solid lines: documented in a cited record. Dashed: reported or inferred. Hollow arrows: proposed powers. Choose a connection for its evidence.
Coordinator accepted / Ready for submission is an internal designation, not Commission acceptance. Keywords select material for research; they do not establish truth, illegality or deployed enforcement.
Global Disinformation Index → contributed pre-existing keyword resources to the design → AI4TRUST D2.1 methodological toolbox
Submitted1November2023; individual contribution dates unknown.
D2.1 explicitly uses existing GDI lists among research, media and fact-checker contributions and some ChatGPT suggestions. No individual seed term is attributed exclusively to GDI, and the report does not establish which legal entity owned each background asset.
Fondazione Bruno Kessler: AI4TRUST coordinator → edited and recorded coordinator review before submission → AI4TRUST D2.1 methodological toolbox
2023-11-01 – 2023-11-01
D2.1 identifies FBK as editor and records final coordinator review on 1 November 2023. PDF2 positively designates PU: Public, UCAM reviewer, Coordinator accepted and Ready for submission in bold. These are internal document designations, not Commission acceptance.
UCAM: D2.1 quality-assurance role → is credited with the designated quality-manager review → AI4TRUST D2.1 methodological toolbox
2023-10-27 – 2023-10-27
Modification table version0.6 credits UCAM quality review27October2023. This is internal document review, not a public threshold approval policy or Commission acceptance.
AI4TRUST D2.1 methodological toolbox → defines a mixed-source collection perimeter → AI4TRUST retrieval seeds and sampling frame
Selected initial topics and reusable relevance/feasibility criteria guide development/test collection. Neutral topics coexist with charged language; broad lists are refined. Inclusion is not a person/site falsehood finding, automatic sanction list or proven runtime configuration.
AI4TRUST D2.1 methodological toolbox → expressly adopts an intersectional methodological frame → D2.1 intersectional methodological framing
PDF58-59 describes attention to harm, discrimination and social identity. The effect on annotation labels, thresholds or coverage remains a question for further records, not an observed exclusion result.