diff --git a/src/databricks/sqlalchemy/_types.py b/src/databricks/sqlalchemy/_types.py
index 5fc14a70..a221c7ce 100644
--- a/src/databricks/sqlalchemy/_types.py
+++ b/src/databricks/sqlalchemy/_types.py
@@ -33,6 +33,7 @@ def process_literal_param_hack(value: Any):
 @compiles(sqlalchemy.types.Unicode, "databricks")
 @compiles(sqlalchemy.types.UnicodeText, "databricks")
 @compiles(sqlalchemy.types.Uuid, "databricks")
+@compiles(sqlalchemy.types.JSON, "databricks")
 def compile_string_databricks(type_, compiler, **kw):
     """
     We override the default compilation for Enum(), String(), Text(), and Time() because SQLAlchemy
diff --git a/src/databricks/sqlalchemy/test_local/test_types.py b/src/databricks/sqlalchemy/test_local/test_types.py
index b91217ed..a843d93d 100644
--- a/src/databricks/sqlalchemy/test_local/test_types.py
+++ b/src/databricks/sqlalchemy/test_local/test_types.py
@@ -128,6 +128,7 @@ def test_numeric_renders_as_decimal_with_precision_and_scale(self):
     sqlalchemy.types.INT: DatabricksDataType.INT,
     sqlalchemy.types.SMALLINT: DatabricksDataType.SMALLINT,
     sqlalchemy.types.TIMESTAMP: DatabricksDataType.TIMESTAMP,
+    sqlalchemy.types.JSON: DatabricksDataType.STRING,
     TINYINT: DatabricksDataType.TINYINT,
     TIMESTAMP: DatabricksDataType.TIMESTAMP,
     TIMESTAMP_NTZ: DatabricksDataType.TIMESTAMP_NTZ,