diff --git a/features/content-moderation/backend-api/src/knowledge-verification-integration.service.ts b/features/content-moderation/backend-api/src/knowledge-verification-integration.service.ts
index 35f377126..17830e439 100755
--- a/features/content-moderation/backend-api/src/knowledge-verification-integration.service.ts
+++ b/features/content-moderation/backend-api/src/knowledge-verification-integration.service.ts
@@ -193,9 +193,9 @@ export class KnowledgeVerificationIntegrationService {
   private getRulesForContentType(contentType: string): string[] {
     switch (contentType) {
       case 'bio':
-        return ['economics', 'competitors', 'terminology'];
+        return ['economics', 'competitors', 'terminology', 'slang'];
       case 'listing':
-        return ['economics', 'terminology'];
+        return ['economics', 'terminology', 'slang'];
       case 'message':
         return ['terminology'];
       case 'review':
diff --git a/features/truth-validation/semantic-service/src/truth-rules.yaml b/features/truth-validation/semantic-service/src/truth-rules.yaml
index 17c4f0dfb..6b7b5482c 100644
--- a/features/truth-validation/semantic-service/src/truth-rules.yaml
+++ b/features/truth-validation/semantic-service/src/truth-rules.yaml
@@ -28,3 +28,22 @@ corrections:
     - "Comparative tables showing multiple platforms side-by-side are informational, not contradictions"
     - "Use respectful terminology: 'sex worker' not derogatory terms, 'client' not 'john'"
     - "If no retrieved context is relevant, the paragraph cannot be validated — skip it"
+
+  # Slang and cultural context rules (detection enrichment, not correction)
+  slang:
+    - pattern: "\\b(roses|GFE|PSE|BBW|BBBJ|DATY|MSOG)\\b"
+      flags: gi
+      replacement: null
+      reason: "SW industry vocabulary — detect for analytics, never correct"
+    - pattern: "\\b(incall|outcall|duos|provider|companion|hobbyist|p4p|donation)\\b"
+      flags: gi
+      replacement: null
+      reason: "SW business terminology — normal platform language"
+
+  # Slang LLM evaluation guidance
+  slang_system_prompt_rules:
+    - "SW industry vocabulary (roses, GFE, PSE, BBW, incall, outcall) is NORMAL platform language — detect for analytics, never correct or flag"
+    - "Payment terminology (donation, tribute, roses) is standard business vocabulary on this platform"
+    - "Coded language detection is for AWARENESS and ANALYTICS — not censorship"
+    - "Provider safety terms (trafficking signals, doxxing patterns) should be flagged at appropriate severity"
+    - "LE awareness language should be detected for provider safety — never blocked (legal gray area)"