diff --git a/backend/open_webui/routers/audio.py b/backend/open_webui/routers/audio.py
index 27634cec199f..4f310f598522 100644
--- a/backend/open_webui/routers/audio.py
+++ b/backend/open_webui/routers/audio.py
@@ -49,6 +49,7 @@
     DEVICE_TYPE,
     ENABLE_FORWARD_USER_INFO_HEADERS,
 )
+from open_webui.utils.content_types import get_stt_supported_content_types
 
 
 router = APIRouter()
@@ -238,9 +239,11 @@ async def update_audio_config(
     request.app.state.config.STT_OPENAI_API_KEY = form_data.stt.OPENAI_API_KEY
     request.app.state.config.STT_ENGINE = form_data.stt.ENGINE
     request.app.state.config.STT_MODEL = form_data.stt.MODEL
-    request.app.state.config.STT_SUPPORTED_CONTENT_TYPES = (
-        form_data.stt.SUPPORTED_CONTENT_TYPES
-    )
+    request.app.state.config.STT_SUPPORTED_CONTENT_TYPES = [
+        content_type.strip()
+        for content_type in form_data.stt.SUPPORTED_CONTENT_TYPES
+        if content_type.strip()
+    ]
 
     request.app.state.config.WHISPER_MODEL = form_data.stt.WHISPER_MODEL
     request.app.state.config.DEEPGRAM_API_KEY = form_data.stt.DEEPGRAM_API_KEY
@@ -919,10 +922,9 @@ def transcription(
 ):
     log.info(f"file.content_type: {file.content_type}")
 
-    supported_content_types = request.app.state.config.STT_SUPPORTED_CONTENT_TYPES or [
-        "audio/*",
-        "video/webm",
-    ]
+    supported_content_types = get_stt_supported_content_types(
+        request.app.state.config.STT_SUPPORTED_CONTENT_TYPES
+    )
 
     if not any(
         fnmatch(file.content_type, content_type)
diff --git a/backend/open_webui/routers/files.py b/backend/open_webui/routers/files.py
index b9bb15c7b400..7b76cb5bbd32 100644
--- a/backend/open_webui/routers/files.py
+++ b/backend/open_webui/routers/files.py
@@ -36,6 +36,7 @@
 from open_webui.routers.audio import transcribe
 from open_webui.storage.provider import Storage
 from open_webui.utils.auth import get_admin_user, get_verified_user
+from open_webui.utils.content_types import get_stt_supported_content_types
 from pydantic import BaseModel
 
 log = logging.getLogger(__name__)
@@ -155,12 +156,8 @@ def upload_file(
         if process:
             try:
                 if file.content_type:
-                    stt_supported_content_types = (
+                    stt_supported_content_types = get_stt_supported_content_types(
                         request.app.state.config.STT_SUPPORTED_CONTENT_TYPES
-                        or [
-                            "audio/*",
-                            "video/webm",
-                        ]
                     )
 
                     if any(
diff --git a/backend/open_webui/routers/ollama.py b/backend/open_webui/routers/ollama.py
index 135359937447..2c7d4119b258 100644
--- a/backend/open_webui/routers/ollama.py
+++ b/backend/open_webui/routers/ollama.py
@@ -48,8 +48,9 @@
     apply_model_params_to_body_openai,
     apply_model_system_prompt_to_body,
 )
+from open_webui.utils.messages import remove_user_system_messages_from_body
 from open_webui.utils.auth import get_admin_user, get_verified_user
-from open_webui.utils.access_control import has_access
+from open_webui.utils.access_control import has_access, has_permission
 
 
 from open_webui.config import (
@@ -69,6 +70,17 @@
 log.setLevel(SRC_LOG_LEVELS["OLLAMA"])
 
 
+def user_can_use_system_prompt(request: Request, user: UserModel) -> bool:
+    if user.role == "admin":
+        return True
+
+    return has_permission(
+        user.id,
+        "chat.system_prompt",
+        request.app.state.config.USER_PERMISSIONS,
+    )
+
+
 ##########################################
 #
 # Utility functions
@@ -1289,6 +1301,9 @@ async def generate_chat_completion(
     if "metadata" in payload:
         del payload["metadata"]
 
+    if not user_can_use_system_prompt(request, user):
+        payload = remove_user_system_messages_from_body(payload)
+
     model_id = payload["model"]
     model_info = Models.get_model_by_id(model_id)
 
@@ -1472,6 +1487,9 @@ async def generate_openai_chat_completion(
     if "metadata" in payload:
         del payload["metadata"]
 
+    if not user_can_use_system_prompt(request, user):
+        payload = remove_user_system_messages_from_body(payload)
+
     model_id = completion_form.model
     if ":" not in model_id:
         model_id = f"{model_id}:latest"
diff --git a/backend/open_webui/routers/openai.py b/backend/open_webui/routers/openai.py
index 7649271feebb..2a29b5cb7ef4 100644
--- a/backend/open_webui/routers/openai.py
+++ b/backend/open_webui/routers/openai.py
@@ -37,18 +37,30 @@
     apply_model_params_to_body_openai,
     apply_model_system_prompt_to_body,
 )
+from open_webui.utils.messages import remove_user_system_messages_from_body
 from open_webui.utils.misc import (
     convert_logit_bias_input_to_json,
 )
 
 from open_webui.utils.auth import get_admin_user, get_verified_user
-from open_webui.utils.access_control import has_access
+from open_webui.utils.access_control import has_access, has_permission
 
 
 log = logging.getLogger(__name__)
 log.setLevel(SRC_LOG_LEVELS["OPENAI"])
 
 
+def user_can_use_system_prompt(request: Request, user: UserModel) -> bool:
+    if user.role == "admin":
+        return True
+
+    return has_permission(
+        user.id,
+        "chat.system_prompt",
+        request.app.state.config.USER_PERMISSIONS,
+    )
+
+
 ##########################################
 #
 # Utility functions
@@ -705,6 +717,9 @@ async def generate_chat_completion(
     payload = {**form_data}
     metadata = payload.pop("metadata", None)
 
+    if not user_can_use_system_prompt(request, user):
+        payload = remove_user_system_messages_from_body(payload)
+
     model_id = form_data.get("model")
     model_info = Models.get_model_by_id(model_id)
 
diff --git a/backend/open_webui/routers/users.py b/backend/open_webui/routers/users.py
index 16cc2c375f95..58fc328c11c3 100644
--- a/backend/open_webui/routers/users.py
+++ b/backend/open_webui/routers/users.py
@@ -215,9 +215,10 @@ async def update_user_settings_by_session_user(
     request: Request, form_data: UserSettings, user=Depends(get_verified_user)
 ):
     updated_user_settings = form_data.model_dump()
+    ui_settings = updated_user_settings.get("ui") or {}
     if (
         user.role != "admin"
-        and "toolServers" in updated_user_settings.get("ui").keys()
+        and "toolServers" in ui_settings.keys()
         and not has_permission(
             user.id,
             "features.direct_tool_servers",
@@ -227,6 +228,15 @@ async def update_user_settings_by_session_user(
         # If the user is not an admin and does not have permission to use tool servers, remove the key
         updated_user_settings["ui"].pop("toolServers", None)
 
+    if user.role != "admin" and not has_permission(
+        user.id,
+        "chat.system_prompt",
+        request.app.state.config.USER_PERMISSIONS,
+    ):
+        updated_user_settings.pop("system", None)
+        if isinstance(updated_user_settings.get("params"), dict):
+            updated_user_settings["params"].pop("system", None)
+
     user = Users.update_user_settings_by_id(user.id, updated_user_settings)
     if user:
         return user.settings
diff --git a/backend/open_webui/utils/chat.py b/backend/open_webui/utils/chat.py
index 268c910e3ec2..5a4bdd72f863 100644
--- a/backend/open_webui/utils/chat.py
+++ b/backend/open_webui/utils/chat.py
@@ -45,7 +45,11 @@
     get_function_module_from_cache,
 )
 from open_webui.utils.models import get_all_models, check_model_access
-from open_webui.utils.payload import convert_payload_openai_to_ollama
+from open_webui.utils.payload import (
+    convert_payload_openai_to_ollama,
+)
+from open_webui.utils.messages import remove_user_system_messages_from_body
+from open_webui.utils.access_control import has_permission
 from open_webui.utils.response import (
     convert_response_ollama_to_openai,
     convert_streaming_response_ollama_to_openai,
@@ -63,6 +67,17 @@
 log.setLevel(SRC_LOG_LEVELS["MAIN"])
 
 
+def user_can_use_system_prompt(request: Request, user: Any) -> bool:
+    if user.role == "admin":
+        return True
+
+    return has_permission(
+        user.id,
+        "chat.system_prompt",
+        request.app.state.config.USER_PERMISSIONS,
+    )
+
+
 async def generate_direct_chat_completion(
     request: Request,
     form_data: dict,
@@ -189,6 +204,9 @@ async def generate_chat_completion(
     if model_id not in models:
         raise Exception("Model not found")
 
+    if not user_can_use_system_prompt(request, user):
+        form_data = remove_user_system_messages_from_body(form_data)
+
     model = models[model_id]
 
     if getattr(request.state, "direct", False):
diff --git a/backend/open_webui/utils/content_types.py b/backend/open_webui/utils/content_types.py
new file mode 100644
index 000000000000..42f1f9d6869f
--- /dev/null
+++ b/backend/open_webui/utils/content_types.py
@@ -0,0 +1,14 @@
+from typing import Optional
+
+
+DEFAULT_STT_SUPPORTED_CONTENT_TYPES = ["audio/*", "video/webm"]
+
+
+def get_stt_supported_content_types(configured_content_types: Optional[list[str]]):
+    content_types = [
+        content_type.strip()
+        for content_type in configured_content_types or []
+        if isinstance(content_type, str) and content_type.strip()
+    ]
+
+    return content_types or DEFAULT_STT_SUPPORTED_CONTENT_TYPES
diff --git a/backend/open_webui/utils/messages.py b/backend/open_webui/utils/messages.py
new file mode 100644
index 000000000000..e3d36a73c827
--- /dev/null
+++ b/backend/open_webui/utils/messages.py
@@ -0,0 +1,13 @@
+def remove_user_system_messages_from_body(form_data: dict) -> dict:
+    messages = form_data.get("messages")
+    if isinstance(messages, list):
+        form_data["messages"] = [
+            message
+            for message in messages
+            if not (
+                isinstance(message, dict)
+                and str(message.get("role", "")).lower() == "system"
+            )
+        ]
+
+    return form_data
diff --git a/tests/test_critical_permission_regressions.py b/tests/test_critical_permission_regressions.py
new file mode 100644
index 000000000000..a80c65ed24d0
--- /dev/null
+++ b/tests/test_critical_permission_regressions.py
@@ -0,0 +1,49 @@
+import importlib.util
+from pathlib import Path
+
+
+def load_module(module_name: str, relative_path: str):
+    module_path = Path(__file__).parents[1] / relative_path
+    spec = importlib.util.spec_from_file_location(module_name, module_path)
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+content_types = load_module(
+    "content_types", "backend/open_webui/utils/content_types.py"
+)
+messages = load_module("messages", "backend/open_webui/utils/messages.py")
+
+DEFAULT_STT_SUPPORTED_CONTENT_TYPES = content_types.DEFAULT_STT_SUPPORTED_CONTENT_TYPES
+get_stt_supported_content_types = content_types.get_stt_supported_content_types
+remove_user_system_messages_from_body = messages.remove_user_system_messages_from_body
+
+
+def test_blank_stt_content_types_fall_back_to_defaults():
+    assert get_stt_supported_content_types(["", "  "]) == DEFAULT_STT_SUPPORTED_CONTENT_TYPES
+
+
+def test_stt_content_types_are_trimmed_and_empty_entries_removed():
+    assert get_stt_supported_content_types([" audio/wav ", "", "video/mp4"]) == [
+        "audio/wav",
+        "video/mp4",
+    ]
+
+
+def test_user_system_messages_are_removed_from_payload():
+    payload = {
+        "messages": [
+            {"role": "system", "content": "user controlled"},
+            {"role": "user", "content": "hello"},
+            {"role": "SYSTEM", "content": "also user controlled"},
+            {"role": "assistant", "content": "hi"},
+        ]
+    }
+
+    assert remove_user_system_messages_from_body(payload) == {
+        "messages": [
+            {"role": "user", "content": "hello"},
+            {"role": "assistant", "content": "hi"},
+        ]
+    }