Continuous AI-based biodiversity assessments for Germany (KIBA-D)

The project addresses the challenge that a continuous biodiversity assessment (‘Biodiversity Fact-Check’) must efficiently analyse and continuously expand large volumes of heterogeneous, and in some cases legally sensitive, text and data sources. To meet these scaling and timeliness requirements, AI/LLM-supported workflows are to facilitate the (partially) automated recreation and updating of key content. This development task will be carried out centrally by participating researchers specialising in artificial intelligence, as well as experts in the field of biodiversity.

The contribution to be carried out by the Öko-Institut and UfU e.V. (WP 8 “Legal Review and Validation”) focuses on integrated legal support across all work packages (“compliance by design”). To this end, relevant requirements – including those from copyright law, EU and national data law (e.g. Data Act, Data Governance Act, DSA, GDPR) – as well as AI regulatory guidelines will be examined and translated into practical guidelines. In parallel, a data governance concept is being developed (roles/responsibilities, decision-making and documentation processes, criteria for legally compliant data use and data quality). In this way, the legal support for the project not only clarifies the prerequisites for legal admissibility but also establishes an integrated and dynamic governance framework to ensure the legal trustworthiness of AI systems.