fix(pii): include ORGANIZATION in app-side NER set (align with server)

Author: TheodoreSpeaks

apps/sim/lib/guardrails/pii-entities.test.ts

@@ -10,8 +10,10 @@ import {
 } from '@/lib/guardrails/pii-entities'
 
 describe('NER_PII_ENTITIES', () => {
-  it('is exactly the 4 spaCy-NER entities', () => {
-    expect(new Set(NER_PII_ENTITIES)).toEqual(new Set(['PERSON', 'LOCATION', 'NRP', 'DATE_TIME']))
+  it('covers the spaCy-NER entities including ORGANIZATION', () => {
+    expect(new Set(NER_PII_ENTITIES)).toEqual(
+      new Set(['PERSON', 'LOCATION', 'NRP', 'DATE_TIME', 'ORGANIZATION'])
+    )
   })
 })
 
@@ -23,6 +25,7 @@ describe('stripNerEntities', () => {
         'EMAIL_ADDRESS',
         'DATE_TIME',
         'US_SSN',
+        'ORGANIZATION',
         'LOCATION',
         'PHONE_NUMBER',
       ])

apps/sim/lib/guardrails/pii-entities.ts

@@ -230,22 +230,26 @@ export function isEntitySupportedForLanguage(
 }
 
 /**
- * The spaCy-NER entity types — the only ones detected by the NLP model, vs the
- * regex/checksum pattern recognizers. The block-output redaction stage is
- * restricted to the non-NER (regex) entities so it runs on the Presidio
- * spaCy-free fast path without the per-leaf NER cost. Mirrors `NER_ENTITIES` in
- * `apps/pii/server.py`.
+ * Entity types produced by the spaCy NER model (vs the regex/checksum pattern
+ * recognizers). The block-output redaction stage is restricted to the non-NER
+ * (regex) entities so it runs on the Presidio spaCy-free fast path without the
+ * per-leaf NER cost. Includes ORGANIZATION — which Presidio's spaCy recognizer
+ * emits but the user-facing catalog above does not list — so it too is stripped
+ * from block-output selections, keeping this in sync with the derived
+ * `NER_ENTITIES` in `apps/pii/server.py`. Typed as strings because ORGANIZATION
+ * isn't a catalog `PIIEntityType`.
  */
-export const NER_PII_ENTITIES: ReadonlySet<PIIEntityType> = new Set<PIIEntityType>([
+export const NER_PII_ENTITIES: ReadonlySet<string> = new Set<string>([
   'PERSON',
   'LOCATION',
   'NRP',
   'DATE_TIME',
+  'ORGANIZATION',
 ])
 
 /** Drop the spaCy-NER entities ({@link NER_PII_ENTITIES}) from a selection. */
 export function stripNerEntities(entities: readonly string[]): string[] {
-  return entities.filter((e) => !NER_PII_ENTITIES.has(e as PIIEntityType))
+  return entities.filter((e) => !NER_PII_ENTITIES.has(e))
 }
 
 /**