Policy: GCP > Model Armor > Floor Setting > Settings > Responsible AI

Parent for the per-Responsible-AI-filter sub-policies of GCP > Model Armor > Floor Setting > Settings. The four detectors here mirror the Responsible AI section of the Model Armor Console floor settings page (Hate Speech, Dangerous, Sexually Explicit, Harassment). The parent has no enforcement value of its own — it groups the leaf policies under one tree node so the Console UI can render them together.

Each leaf takes a confidence level (None / Low and above / Medium and above / High) that maps to the filterConfig.raiSettings.raiFilters[] entry on the live Floor Setting. None means "do not require this RAI filter"; the others map to the corresponding confidenceLevel enum at the API.