diff --git a/litellm/proxy/guardrails/guardrail_hooks/trendai/__init__.py b/litellm/proxy/guardrails/guardrail_hooks/trendai/__init__.py
index b6ca0d3a7c7..7803ac0baea 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/trendai/__init__.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/trendai/__init__.py
@@ -28,7 +28,6 @@ def initialize_guardrail(litellm_params: "LitellmParams", guardrail: "Guardrail"
         api_base=litellm_params.api_base,
         app_name=settings.app_name,
         fallback_on_error=settings.fallback_on_error,
-        mask_pii=settings.mask_pii,
         timeout=settings.timeout,
         stream_batch_size=settings.stream_batch_size,
         stream_overlap_size=settings.stream_overlap_size,
diff --git a/litellm/proxy/guardrails/guardrail_hooks/trendai/_models.py b/litellm/proxy/guardrails/guardrail_hooks/trendai/_models.py
index 2efcbb52c75..0acaf1fb0d3 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/trendai/_models.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/trendai/_models.py
@@ -11,7 +11,6 @@ from pydantic import BaseModel, ConfigDict, Field
 class TrendAISettings(BaseModel):
     app_name: str | None = None
     fallback_on_error: Literal["block", "allow"] = "block"
-    mask_pii: bool = True
     timeout: float = 5.0
     stream_batch_size: int = 2048
     stream_overlap_size: int = 256
diff --git a/litellm/proxy/guardrails/guardrail_hooks/trendai/trendai.py b/litellm/proxy/guardrails/guardrail_hooks/trendai/trendai.py
index 4ed0e54abe8..c6753738ffc 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/trendai/trendai.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/trendai/trendai.py
@@ -74,7 +74,6 @@ class TrendAIGuardrail(CustomGuardrail):
         api_base: str | None = None,
         app_name: str | None = None,
         fallback_on_error: Literal["block", "allow"] = "block",
-        mask_pii: bool = True,
         timeout: float = 5.0,
         stream_batch_size: int = 2048,
         stream_overlap_size: int = 256,
@@ -114,7 +113,6 @@ class TrendAIGuardrail(CustomGuardrail):
         self.api_url: str = _build_apply_guardrails_url(resolved_api_base)
         self.app_name: str = app_name or os.environ.get("TMV1_APPLICATION_NAME", "litellm")
         self.fallback_on_error: Literal["block", "allow"] = fallback_on_error
-        self.mask_pii: bool = mask_pii
         self.timeout: float = timeout
         self.stream_batch_size: int = stream_batch_size
         self.stream_overlap_size: int = stream_overlap_size
@@ -129,8 +127,8 @@ class TrendAIGuardrail(CustomGuardrail):
             event_hook=event_hook,
             default_on=default_on,
             supported_event_hooks=self.get_supported_event_hooks(),
-            mask_request_content=mask_pii,
-            mask_response_content=mask_pii,
+            mask_request_content=True,
+            mask_response_content=True,
         )
 
     @classmethod
@@ -369,7 +367,7 @@ class TrendAIGuardrail(CustomGuardrail):
             ("TMV1-Client-Name", TMV1_CLIENT_NAME),
             ("TMV1-Client-Version", litellm_version),
             ("TMV1-Plugin-Version", PLUGIN_VERSION),
-            *_optional_header("prefer", "redact-pii,return=representation" if self.mask_pii else None),
+            ("prefer", "redact-pii,return=representation"),
             *_optional_header("TMV1-Request-Type", request_type),
         )
         return MappingProxyType(dict(headers))
diff --git a/tests/test_litellm/proxy/guardrails/guardrail_hooks/trendai/test_trendai.py b/tests/test_litellm/proxy/guardrails/guardrail_hooks/trendai/test_trendai.py
index 65b93c53d68..4ae752cecd4 100644
--- a/tests/test_litellm/proxy/guardrails/guardrail_hooks/trendai/test_trendai.py
+++ b/tests/test_litellm/proxy/guardrails/guardrail_hooks/trendai/test_trendai.py
@@ -33,7 +33,6 @@ def _guardrail(
     *,
     app_name: str | None = None,
     fallback_on_error: Literal["block", "allow"] = "block",
-    mask_pii: bool = True,
     timeout: float = 5.0,
     stream_batch_size: int = 2048,
     stream_overlap_size: int = 256,
@@ -48,7 +47,6 @@ def _guardrail(
         api_base=api_base,
         app_name=app_name,
         fallback_on_error=fallback_on_error,
-        mask_pii=mask_pii,
         timeout=timeout,
         stream_batch_size=stream_batch_size,
         stream_overlap_size=stream_overlap_size,
@@ -133,6 +131,15 @@ def test_invalid_timeout_is_rejected() -> None:
         _guardrail(timeout=0)
 
 
+def test_pii_masking_is_not_configurable_on_the_guardrail() -> None:
+    with pytest.raises(TypeError, match="mask_pii"):
+        TrendAIGuardrail(
+            api_key="test-key",
+            api_base="https://guard.example.com/v3.0/aiSecurity",
+            mask_pii=False,
+        )
+
+
 @pytest.mark.asyncio
 async def test_scan_uses_injected_client_and_required_headers() -> None:
     captured_request: httpx.Request | None = None