diff --git a/CHANGELOG.md b/CHANGELOG.md
index 07ab849b..2833ba66 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -26,6 +26,8 @@
 - Added deterministic semantic rounding for model context and normalized model
   Markdown to the existing plain-text renderer so hectare, population and
   score answers remain readable without exposing raw floating-point tails.
+- Prevented an area-weighted population estimate from retaining model phrases
+  such as "official count" when its persisted metric is marked as estimated.
 
 ## Sprint 213-214 Cross-domain area profile (2026-07-16)
 
diff --git a/backend/app/services/geo_assistant_service.py b/backend/app/services/geo_assistant_service.py
index 85f843c8..2fd867f5 100644
--- a/backend/app/services/geo_assistant_service.py
+++ b/backend/app/services/geo_assistant_service.py
@@ -108,13 +108,27 @@ class GeoAssistantService:
         answer: str,
         metrics: list[AssistantContextMetric],
     ) -> str:
+        estimated_themes = {metric.theme for metric in metrics if metric.is_estimate}
+        if "population" in estimated_themes:
+            answer = re.sub(
+                r"\bde officiële telling\b",
+                "de van de officiële bron afgeleide schatting",
+                answer,
+                flags=re.IGNORECASE,
+            )
+            answer = re.sub(
+                r"\bofficieel geteld aantal inwoners\b",
+                "uit een officiële bron afgeleid aantal inwoners",
+                answer,
+                flags=re.IGNORECASE,
+            )
         normalized = answer.casefold()
         if "schat" in normalized:
             return answer
         disclosed_themes = {
             metric.theme
             for metric in metrics
-            if metric.is_estimate
+            if metric.theme in estimated_themes
             and any(
                 term in normalized
                 for term in cls.ESTIMATE_TOPIC_TERMS.get(metric.theme, (metric.label.casefold(),))
diff --git a/backend/tests/test_sprint202_temporal_metrics_and_ollama.py b/backend/tests/test_sprint202_temporal_metrics_and_ollama.py
index 17231571..8c604bb5 100644
--- a/backend/tests/test_sprint202_temporal_metrics_and_ollama.py
+++ b/backend/tests/test_sprint202_temporal_metrics_and_ollama.py
@@ -144,6 +144,28 @@ def test_geo_assistant_discloses_estimated_population_values() -> None:
     )
 
 
+def test_geo_assistant_never_labels_area_weighted_population_as_official_count() -> None:
+    metrics = [
+        AssistantContextMetric(
+            theme="population",
+            label="Geraamd aantal inwoners",
+            value=38_675,
+            unit="inwoners",
+            source="Statbel",
+            dataset_id=uuid4(),
+            is_estimate=True,
+        )
+    ]
+
+    answer = GeoAssistantService.ensure_estimate_disclosure(
+        "De officiële telling uit januari 2025 bedraagt 38.675 inwoners. Deze waarde is een schatting.",
+        metrics,
+    )
+
+    assert "officiële telling" not in answer.casefold()
+    assert answer.startswith("de van de officiële bron afgeleide schatting")
+
+
 def test_geo_assistant_does_not_add_irrelevant_estimate_disclosure() -> None:
     metrics = [
         AssistantContextMetric(
diff --git a/docs/CODEX_EXECUTION_LOG.md b/docs/CODEX_EXECUTION_LOG.md
index b303a8ef..077919b7 100644
--- a/docs/CODEX_EXECUTION_LOG.md
+++ b/docs/CODEX_EXECUTION_LOG.md
@@ -20,6 +20,10 @@ Changed:
   Markdown and raw floating-point tails in the plain-text chat renderer. Added
   deterministic Markdown normalization and unit-aware context rounding while
   retaining the unrounded metrics in the canonical API response.
+- The next live answer still called an area-weighted Statbel population value
+  an official count before acknowledging it as an estimate. The deterministic
+  estimate guard now removes that contradictory label whenever persisted
+  population context is estimated.
 
 Validation evidence:
 - A read-only live production-chain probe with the 1,200-token setting
@@ -33,6 +37,8 @@ Validation evidence:
   shell gates.
 - After plain-text normalization and semantic prompt rounding, the final
   readiness rerun passed 725 backend tests and the same complete gate set.
+- After hardening estimated-population wording, the final complete readiness
+  rerun passed 726 backend tests and all remaining gates unchanged.
 - The regional thematic operator dry-run resolved exactly 28 official
   municipality Areas. The live Mol run in `Kempen Regional Workbench` imported
   all five products with complete source coverage; the DOV operator imported