diff --git a/src/clusterfuzz/_internal/base/utils.py b/src/clusterfuzz/_internal/base/utils.py
index 55b208d7fb..43a972c851 100644
--- a/src/clusterfuzz/_internal/base/utils.py
+++ b/src/clusterfuzz/_internal/base/utils.py
@@ -103,7 +103,7 @@ def decode_to_unicode(obj):
   if not hasattr(obj, 'decode'):
     return obj
 
-  return obj.decode('utf-8', errors='ignore')
+  return obj.decode('utf-8', errors='replace')
 
 
 def encode_as_unicode(obj):
diff --git a/src/clusterfuzz/_internal/bot/tasks/utasks/fuzz_task.py b/src/clusterfuzz/_internal/bot/tasks/utasks/fuzz_task.py
index 419590bc59..e3e585c38f 100644
--- a/src/clusterfuzz/_internal/bot/tasks/utasks/fuzz_task.py
+++ b/src/clusterfuzz/_internal/bot/tasks/utasks/fuzz_task.py
@@ -760,6 +760,9 @@ def add_additional_testcase_run_data(testcase_run, fully_qualified_fuzzer_name,
 def get_fuzzer_metadata_from_output(fuzzer_output):
   """Extract metadata from fuzzer output."""
   metadata = {}
+  fuzzer_output = fuzzer_output.replace(
+      '\r', '').replace('\x85', '').replace(
+      '\u2028', '').replace('\u2029', '')
   for line in fuzzer_output.splitlines():
     match = FUZZER_METADATA_REGEX.match(line)
     if match: