diff --git a/src/diffusers/utils/loading_utils.py b/src/diffusers/utils/loading_utils.py
index ba528c7f857c..9d729d31157c 100644
--- a/src/diffusers/utils/loading_utils.py
+++ b/src/diffusers/utils/loading_utils.py
@@ -143,6 +143,8 @@ def load_video(
 
     if convert_method is not None:
         pil_images = convert_method(pil_images)
+    else:
+        pil_images = [image.convert("RGB") for image in pil_images]
 
     if return_fps:
         if fps is None:
diff --git a/tests/others/test_utils.py b/tests/others/test_utils.py
index 0341623d8783..a9df31befdf1 100755
--- a/tests/others/test_utils.py
+++ b/tests/others/test_utils.py
@@ -17,11 +17,12 @@
 import os
 import warnings
 
+import PIL.Image
 import pytest
 import torch
 
 from diffusers import __version__
-from diffusers.utils import deprecate, torch_utils
+from diffusers.utils import deprecate, load_video, torch_utils
 from diffusers.utils.torch_utils import TorchDeviceBackend, empty_device_cache, get_device
 
 from ..testing_utils import CaptureLogger, Expectations, str_to_bool
@@ -372,6 +373,18 @@ def check(value, key):
             expectations.find_expectation(("xpu", None))
 
 
+class TestLoadVideo:
+    def test_gif_frames_are_rgb(self, tmp_path):
+        # Pillow decodes the first GIF frame as "P" and later frames as "RGB"
+        frames = [PIL.Image.new("RGB", (8, 8), color) for color in ("red", "green", "blue")]
+        path = str(tmp_path / "video.gif")
+        frames[0].save(path, save_all=True, append_images=frames[1:], duration=100)
+
+        video = load_video(path)
+
+        assert [frame.mode for frame in video] == ["RGB", "RGB", "RGB"]
+
+
 def parse_flag_from_env(key, default=False):
     try:
         value = os.environ[key]