diff --git a/core/database_arango.py b/core/database_arango.py
index 83dbac3b9..90edf033d 100644
--- a/core/database_arango.py
+++ b/core/database_arango.py
@@ -2,17 +2,13 @@
 import datetime
 import json
 import logging
-import re
 import sys
 import time
-import unicodedata
 from typing import TYPE_CHECKING, Any, Iterable, List, Optional, Tuple, Type, TypeVar
 
 if TYPE_CHECKING:
+    from core.schemas import entity, indicator, observable
     from core.schemas.graph import Relationship, TagRelationship
-    from core.schemas import observable
-    from core.schemas import entity
-    from core.schemas import indicator
 
 import requests
 from arango import ArangoClient
@@ -303,41 +299,34 @@ def tag(
     ) -> TYetiObject:
         """Connects object to tag graph."""
         # Import at runtime to avoid circular dependency.
-        from core.schemas.tag import DEFAULT_EXPIRATION_DAYS, Tag
+        from core.schemas import tag
 
-        expiration_days = expiration_days or DEFAULT_EXPIRATION_DAYS
+        expiration_days = expiration_days or tag.DEFAULT_EXPIRATION_DAYS
 
         if strict:
             self.clear_tags()
 
         extra_tags = set()
-        for tag_name in tags:
-            # Attempt to find replacement tag
-            if normalized:
-                nfkd_form = unicodedata.normalize("NFKD", tag_name)
-                nfkd_form.encode("ASCII", "ignore").decode("UTF-8")
-                tag_name = "".join(
-                    [c for c in nfkd_form if not unicodedata.combining(c)]
-                )
-                tag_name = tag_name.strip()
-                tag_name = re.sub(r"\s+", "_", tag_name).lower()
-                tag_name = re.sub(r"[^a-zA-Z0-9_]", "", tag_name)
-            replacements, _ = Tag.filter({"in__replaces": [tag_name]}, count=1)
-            tag: Optional[Tag] = None
+        for provided_tag_name in tags:
+            tag_name = tag.normalize_name(provided_tag_name)
+            if not tag_name:
+                raise RuntimeError(f"Cannot tag object with empty tag: '{provided_tag_name}' -> '{tag_name}'")
+            replacements, _ = tag.Tag.filter({"in__replaces": [tag_name]}, count=1)
+            new_tag: Optional[tag.Tag] = None
 
             if replacements:
-                tag = replacements[0]
+                new_tag = replacements[0]
             # Attempt to find actual tag
             else:
-                tag = Tag.find(name=tag_name)
+                new_tag = tag.Tag.find(name=tag_name)
             # Create tag
-            if not tag:
-                tag = Tag(name=tag_name).save()
+            if not new_tag:
+                new_tag = tag.Tag(name=tag_name).save()
 
-            tag_link = self.link_to_tag(tag.name)
-            self.tags[tag.name] = tag_link
+            tag_link = self.link_to_tag(new_tag.name)
+            self.tags[new_tag.name] = tag_link
 
-            extra_tags |= set(tag.produces)
+            extra_tags |= set(new_tag.produces)
 
         extra_tags -= set(tags)
         if extra_tags:
diff --git a/core/schemas/tag.py b/core/schemas/tag.py
index 60c1a604a..c6368c111 100644
--- a/core/schemas/tag.py
+++ b/core/schemas/tag.py
@@ -1,4 +1,6 @@
 import datetime
+import re
+import unicodedata
 from typing import ClassVar
 
 from pydantic import BaseModel, Field
@@ -11,6 +13,16 @@
 def future():
     return datetime.timedelta(days=DEFAULT_EXPIRATION_DAYS)
 
+def normalize_name(tag_name: str) -> str:
+    nfkd_form = unicodedata.normalize("NFKD", tag_name)
+    nfkd_form.encode("ASCII", "ignore").decode("UTF-8")
+    tag_name = "".join(
+        [c for c in nfkd_form if not unicodedata.combining(c)]
+    )
+    tag_name = tag_name.strip().lower()
+    tag_name = re.sub(r"\s+", "_", tag_name)
+    tag_name = re.sub(r"[^a-zA-Z0-9_:-]", "", tag_name)
+    return tag_name
 
 class Tag(BaseModel, database_arango.ArangoYetiConnector):
     _collection_name: ClassVar[str] = "tags"
diff --git a/tests/schemas/tag.py b/tests/schemas/tag.py
index 2d2c4d832..37ce04da3 100644
--- a/tests/schemas/tag.py
+++ b/tests/schemas/tag.py
@@ -171,18 +171,18 @@ def test_duplicate_name(self):
 
     def test_normalized_tag(self):
         """Tests that a tag can be normalized."""
-        cases = cases = [
+        cases = [
             ("H@ackërS T3st", "hackers_t3st"),
             ("    SpaCesStartEnd  ", "spacesstartend"),
             ("!!Sp3cial##", "sp3cial"),
-            ("Multi    Spaces", "multi_spaces"),
+            ("Multi    Spaces   After", "multi_spaces_after"),
             ("Élévation", "elevation"),
             ("UNDER_score", "under_score"),
             ("mixCaseMix123", "mixcasemix123"),
             ("MïxedÁccénts", "mixedaccents"),
             ("123456", "123456"),
             ("测试chinese", "chinese"),
-            ("", ""),
+            ("type:some-custom-type", "type:some-custom-type")
         ]
 
         for cmp, (tag_non_norm, tag_norm) in enumerate(cases):