docs(architect): weekly KB update — 66 files refreshed (2026-04)

Updated 66 stale knowledge base reference files (10 critical, 56 high)
across all 5 skills using Microsoft Learn MCP research.

Key factual updates:
- Groundedness Detection API: `correction` → `mitigating` param,
  `correctedText` → `correctionText` (breaking change)
- Copilot Studio: GPT-4.1 mini now default (was GPT-4o mini);
  Claude Sonnet 4.5 + Opus 4.5 added (experimental, 200K ctx)
- Agentic Retrieval: still public preview; 50M free tokens/month
- Azure security baselines: "Cognitive Services" → "Foundry Tools"
- Databricks: Delta Live Tables → Lakeflow Spark Declarative Pipelines
- MLflow 3 GenAI: new Feedback/Expectation data model
- Token tracking doc: "Azure OpenAI in Foundry Models through a gateway"
- Agent Registry: Risks column (M365 E7), Graph API (preview)
- Copilot DLP: new Entra AI Admin + Purview Data Security AI Admin roles
- ISO/IEC 42001: scope expanded to M365 Copilot, Foundry, Security Copilot
- Zero Trust: CAE now via Conditional Access, Strict Location Enforcement
- Purview: new Fabric Copilots/agents governance section
- AG-UI HITL: ApprovalRequiredAIFunction (C#), @tool approval_mode (Python)

All files: Last updated → 2026-04, *(Verified MCP 2026-04)* markers added.
Build registry: 1341 URLs from 387 files (+2 new URLs).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

skills/ms-ai-security/references/ai-security-engineering/pii-detection-norwegian-context.md

@@ -1,6 +1,6 @@
 # PII Detection and Masking in Norwegian Text
 
-**Last updated:** 2026-02
+**Last updated:** 2026-04
 **Status:** GA
 **Category:** AI Security Engineering
 
@@ -48,32 +48,44 @@ Azure AI Language tilbyr tre API-varianter for PII-deteksjon:
 
 **Viktig:** Azure detekterer norske fødselsnummer under kategorien `NOIdentityNumber`. Du må spesifisere `language: "no"` for optimal deteksjon.
 
-### Maskeringsstrategier
+### Maskeringsstrategier *(Verified MCP 2026-04)*
 
 Azure AI Language tilbyr fire redaction policies (2025-11-15-preview):
 
 | Policy | Output | Bruksområde |
 |--------|--------|-------------|
-| **CharacterMask** (default) | `Min SSN er ***********` | Standard masking |
+| **CharacterMask** (default) | `Min SSN er ***********` | Standard masking; støtter valgfri `redactionCharacter` (f.eks. `-`) |
 | **EntityMask** | `Min SSN er [NOIdentityNumber_1]` | Logging, debugging |
-| **SyntheticReplacement** | `Min SSN er 12345678901` | Syntetiske testdata |
-| **NoMask** | `Min SSN er 01019912345` | Kun entitetsdeteksjon |
+| **SyntheticReplacement** | `Min SSN er 12345678901` | Syntetiske testdata (tilfeldig valgte erstatningsverdier fra forhåndsdefinert sett) |
+| **NoMask** | `Min SSN er 01019912345` | Kun entitetsdeteksjon, ingen `redactedText` i respons |
 
 **Anbefalt:** `CharacterMask` for produksjon, `EntityMask` for logging (spesifiserer entitetstype), `NoMask` når du kun trenger deteksjon uten redaction.
 
-### Confidence Threshold
+**Ny: Per-entity policy overrides:** Du kan nå spesifisere ulike policies per entitetstype i samme request, med én `defaultRedactionPolicy` og entitetsspesifikke overrides. *(Verified MCP 2026-04)*
 
-Fra 2025-11-15-preview kan du konfigurere `confidenceScoreThreshold` (0.0-1.0):
+**Ny: DisableEntityValidation** (2025-11-15-preview): Mulighet til å deaktivere streng entitetsvalidering (default `false`) for å øke hastighet i scenarioer der validering ikke er nødvendig. *(Verified MCP 2026-04)*
+
+**Ny: EntitySynonyms og ValueExclusionPolicy:** Tilpass PII-tjenesten til organisasjonens vokabular — definer egne synonymer for entitetstyper, og ekskluder spesifikke termer fra deteksjon (f.eks. "politimann", "vitne"). *(Verified MCP 2026-04)*
+
+### Confidence Threshold *(Verified MCP 2026-04)*
+
+Fra 2025-11-15-preview kan du konfigurere `confidenceScoreThreshold` med global default og per-entitet, per-språk overrides:
 
 ```json
 {
   "parameters": {
-    "confidenceScoreThreshold": 0.8
+    "confidenceScoreThreshold": {
+      "default": 0.9,
+      "overrides": [
+        { "value": 0.8, "entity": "NOIdentityNumber" },
+        { "value": 0.6, "entity": "Person", "language": "no" }
+      ]
+    }
   }
 }
 ```
 
-**Råd:** Bruk 0.8+ for produksjon (minimerer false positives), 0.6+ for utviklingsmiljø.
+**Råd:** Bruk 0.8+ global default for produksjon (minimerer false positives), 0.6+ for utviklingsmiljø. Per-entitet overrides gir finkornet kontroll. *(Verified MCP 2026-04)*
 
 ## Arkitekturmønstre
 
@@ -170,7 +182,7 @@ Original data → Azure AI Language PII → Pseudonymisering → Sekundær datab
 
 ## Integrasjon med Microsoft-stakken
 
-### Azure AI Foundry
+### Azure AI Foundry *(Verified MCP 2026-04)*
 
 **Playground:** Test PII-deteksjon i [Azure AI Foundry portal](https://ai.azure.com/):
 1. Naviger til Language → PII Detection
@@ -179,7 +191,9 @@ Original data → Azure AI Language PII → Pseudonymisering → Sekundær datab
 4. Lim inn tekst med fødselsnummer
 5. Se detekterte entiteter med confidence scores
 
-**Model deployment:** Bruk `modelVersion: "latest"` for GA-modellen, `"2025-11-15-preview"` for nye features.
+**Model deployment:** Bruk `modelVersion: "latest"` for GA-modellen, `"2025-11-15-preview"` for nye preview-features.
+
+**Merk:** Microsoft Foundry (new) — ny portal med Foundry-prosjekter — og Foundry (classic) er begge tilgjengelige via `https://ai.azure.com/`. For opprettelse av Language-ressurs, bruk **Azure Language in Foundry Tools**. *(Verified MCP 2026-04)*
 
 ### Copilot Studio
 
@@ -400,12 +414,12 @@ df_masked = df.withColumn("text_masked", mask_pii_udf(df.text))
 
 ## Kilder og verifisering
 
-**Verified (fra Microsoft Learn MCP):**
-- [Azure AI Language PII Detection Overview](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/overview) (2025-11-01 GA)
+**Verified (fra Microsoft Learn MCP, re-verifisert 2026-04):** *(Verified MCP 2026-04)*
+- [Azure AI Language PII Detection Overview](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/overview) — Oppdatert: bruker nå «Azure Language in Foundry Tools»-terminologi; ny Foundry (new) portal
 - [Recognized PII and PHI Entities](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/concepts/entity-categories) (inkluderer NOIdentityNumber)
-- [How to: Redact Text PII](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/how-to/redact-text-pii) (redaction policies 2025-11-15-preview)
-- [Quickstart: Detect PII (Python)](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/quickstart?pivots=programming-language-python) (code samples)
-- [Transparency Note for PII](https://learn.microsoft.com/en-us/azure/ai-foundry/responsible-ai/language-service/transparency-note-personally-identifiable-information) (GDPR compliance)
+- [How to: Redact Text PII](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/how-to/redact-text-pii) — Oppdatert: ny DisableEntityValidation, EntitySynonyms, ValueExclusionPolicy, per-entity confidence threshold overrides (2025-11-15-preview)
+- [Quickstart: Detect PII](https://learn.microsoft.com/en-us/azure/ai-services/language-service/personally-identifiable-information/quickstart) — Quickstart er nå for native document PII; link til text/conversation how-to-guides for tekst-PII
+- [Transparency Note for PII](https://learn.microsoft.com/en-us/azure/ai-foundry/responsible-ai/language-service/transparency-note-personally-identifiable-information) (GDPR compliance, nå under Azure AI Foundry responsible AI)
 
 **Baseline (modellkunnskap):**
 - Norsk fødselsnummer-format (11 siffer, mod11-checksumvalidering)
@@ -413,4 +427,4 @@ df_masked = df.withColumn("text_masked", mask_pii_udf(df.text))
 - Personopplysningsloven (norsk GDPR-implementering)
 - Datatilsynets veiledning om pseudonymisering
 
-**Konfidensnivå:** 95% (Verified via Microsoft Learn MCP, Baseline fra kjent standarder)
+**Konfidensnivå:** 95% (Verified via Microsoft Learn MCP 2026-04, Baseline fra kjente standarder)