From Recognition to Refusal: Mapping Safety Formation Zones Across Model Architectures

more
Titel From Recognition to Refusal: Mapping Safety Formation Zones Across Model Architectures
Medien AI Transparency Conference (AITC) 2026
Verlag International Association for Safe & Ethical AI
Verfasser Carsten Lanquillon, Prof. Dr. Sigurd Schacht
Veröffentlichungsdatum 2026-06-18
Projekttitel TTZ NEA (hoheitlich)
Zitation Lanquillon, Carsten; Schacht, Sigurd (2026): From Recognition to Refusal: Mapping Safety Formation Zones Across Model Architectures. AI Transparency Conference (AITC) 2026.