EUROPA Alignment: A Transparent, Parallel Safety Benchmark for 36 European Languages
Technical report, code, benchmark items and run artifacts for EUROPA Alignment, an open multilingual AI-safety benchmark that applies 16 original safety scenarios to each of the 36 languages prioritized by OpenEuroLLM (576 items), with paired harmful/benign contrasts, critical-dimension score gates and a pinned independent LLM judge. The report evaluates four SFT and DPO checkpoints of a 9B OpenEuroLLM model. The PDF is in paper/europa-alignment.pdf.
Authors
- Birger Moëll (ORCID: https://orcid.org/0000-0002-6529-1211)
Institutions
- AI Sweden (SE)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-18
- DOI
- https://doi.org/10.5281/zenodo.24338446
- Citations
- 1
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- article
- Field-Weighted Citation Impact
- 0.97