diff --git a/CHANGELOG.md b/CHANGELOG.md
index 3b60257..6029d2f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,12 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/) and this p
 - upgrade template
 - clarify Producer and Consumer task documentation (ports, entity serialization, dataset advanced-options reference) and fix minor typos in parameter descriptions
 
+### Fixed
+
+- Consumer and Producer tasks no longer convert non-ASCII characters to unicode escape
+  sequences when wrapping consumed Kafka messages into JSON, or when producing messages
+  from a JSON dataset or from entities
+
 ## [3.5.0] 2026-08-24
 
 ### Changed
diff --git a/cmem_plugin_kafka/kafka_handlers.py b/cmem_plugin_kafka/kafka_handlers.py
index 2e71767..3455353 100644
--- a/cmem_plugin_kafka/kafka_handlers.py
+++ b/cmem_plugin_kafka/kafka_handlers.py
@@ -181,7 +181,7 @@ def _split_data(self, data: Response) -> Generator[KafkaMessage]:
             content = None
             if not tombstone:
                 _content = _message.get("content")
-                content = json.dumps(_content) if _content else None
+                content = json.dumps(_content, ensure_ascii=False) if _content else None
             yield KafkaMessage(key=key, value=content, headers=headers, tombstone=tombstone)
 
     def _aggregate_data(self) -> Generator:
@@ -361,7 +361,7 @@ def _split_data(self, data: Entities) -> Generator[KafkaMessage]:
             for i, path in enumerate(paths):
                 values[path.path] = list(entity.values[i])
             result["entity"] = {"uri": entity.uri, "values": values}
-            kafka_payload = json.dumps(result, indent=4)
+            kafka_payload = json.dumps(result, indent=4, ensure_ascii=False)
             yield KafkaMessage(key=None, value=kafka_payload)
 
     def _aggregate_data(self) -> Entities:
diff --git a/cmem_plugin_kafka/utils.py b/cmem_plugin_kafka/utils.py
index 7375d50..a18aec2 100644
--- a/cmem_plugin_kafka/utils.py
+++ b/cmem_plugin_kafka/utils.py
@@ -337,7 +337,7 @@ def get_message_with_json_wrapper(message: KafkaMessage) -> str:
     msg_with_wrapper = {"message": {"key": message.key, "content": json.loads(message.value)}}
     if message.headers:
         msg_with_wrapper["message"]["headers"] = dict(message.headers.items())
-    return json.dumps(msg_with_wrapper, cls=BytesEncoder)
+    return json.dumps(msg_with_wrapper, cls=BytesEncoder, ensure_ascii=False)
 
 
 def get_kafka_statistics(json_data: str) -> dict:
diff --git a/tests/test_split_data_unicode.py b/tests/test_split_data_unicode.py
new file mode 100644
index 0000000..cb708e0
--- /dev/null
+++ b/tests/test_split_data_unicode.py
@@ -0,0 +1,41 @@
+"""Regression tests for non-ASCII character handling in _split_data().
+
+These exercise KafkaJSONDataHandler and KafkaEntitiesDataHandler directly,
+without a Kafka broker or a Corporate Memory connection, since _split_data()
+only transforms in-memory data into KafkaMessage objects.
+"""
+
+import json
+
+import httpx
+from cmem_plugin_base.dataintegration.entity import Entities, Entity, EntityPath, EntitySchema
+from cmem_plugin_base.dataintegration.plugins import PluginLogger
+
+from cmem_plugin_kafka.kafka_handlers import KafkaEntitiesDataHandler, KafkaJSONDataHandler
+
+
+def test_json_data_handler_split_data_keeps_unicode_characters() -> None:
+    """Test that non-ASCII characters from the source dataset are not escaped"""
+    payload = [{"message": {"key": "1", "content": {"city": "Köln", "name": "Müller"}}}]
+    data = httpx.Response(200, content=json.dumps(payload).encode("utf-8"))
+    handler = KafkaJSONDataHandler(context=None, plugin_logger=PluginLogger())  # type: ignore[arg-type]
+
+    messages = list(handler._split_data(data))  # noqa: SLF001
+
+    assert len(messages) == 1
+    assert "\\u00f6" not in messages[0].value
+    assert "\\u00fc" not in messages[0].value
+    assert json.loads(messages[0].value) == {"city": "Köln", "name": "Müller"}
+
+
+def test_entities_data_handler_split_data_keeps_unicode_characters() -> None:
+    """Test that non-ASCII characters in entity values are not escaped"""
+    schema = EntitySchema(type_uri="urn:x-test", paths=[EntityPath(path="city")])
+    entities = Entities(entities=[Entity(uri="urn:x-1", values=[["Köln"]])], schema=schema)
+    handler = KafkaEntitiesDataHandler(context=None, plugin_logger=PluginLogger())  # type: ignore[arg-type]
+
+    messages = list(handler._split_data(entities))  # noqa: SLF001
+
+    assert len(messages) == 1
+    assert "\\u00f6" not in messages[0].value
+    assert json.loads(messages[0].value)["entity"]["values"]["city"] == ["Köln"]
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 74898e2..2cd4720 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -2,7 +2,20 @@
 
 import json
 
-from cmem_plugin_kafka.utils import get_kafka_statistics
+from cmem_plugin_kafka.utils import (
+    KafkaMessage,
+    get_kafka_statistics,
+    get_message_with_json_wrapper,
+)
+
+
+def test_get_message_with_json_wrapper_keeps_unicode_characters() -> None:
+    """Test that non-ASCII characters in the message value are not escaped"""
+    message = KafkaMessage(key="1", value='{"city": "Köln", "name": "Müller"}')
+    wrapped = get_message_with_json_wrapper(message)
+    assert "\\u00f6" not in wrapped
+    assert "\\u00fc" not in wrapped
+    assert json.loads(wrapped)["message"]["content"] == {"city": "Köln", "name": "Müller"}
 
 
 def test_get_kafka_statistics() -> None: