AIMICI Transparency Framework
A framework designed by aimici and tested on the FRAMES project, focused on bringing transparency and understanding of copyright implications to projects with hybrid AI and human workflows.
Scope of attributes
The framework captures the following scope of attributes:
- WHAT - details of the project and its key phases
- WHO - who is responsible for overseeing and reporting on AI usage
- HOW - the AI use cases and tools used
- WHEN - when tools were used and what decisions were made on which use cases or tools can proceed or be stopped
- WHERE - where AI actions took place across a workflow and what assets were affected
Three levels of records
- Level 1: Credits Statement - for audiences and press. A short, plain-language statement in the credits (one to three sentences) about the production’s AI use.
- Level 2: Production AI card - for industry and stakeholders. A one- to two-page reference detailing AI use cases and tools, risk/visibility/ significance attributes, organised into production stages, with a high-level record of human oversight and the businesses involved. Could be published as a website or hosted document.
- Level 3: Auditable AI record - for commissioner-level governance and legal review. Full record including signed declarations, decisions, a high-level actions log, tool and licensing details, records of consent, and granular asset-level data such as machine-readable AI marks (required by the eu-ai-act).
In practice on FRAMES
- The project produced a reviewed production AI card (Level 2) for an AI-centric production for the first time, and explored how the workflow could feed a version of Level 3 records.
- The credit statement (Level 1) was still in progress at paper time.
- The auditable record evolved through the project; machine-readable AI marks emerged on some tools, but the team mostly relied on manual tagging of human vs AI-driven assets in the Streamline platform.