diff --git a/backend/requirements-runtime.txt b/backend/requirements-runtime.txt
index 8ca09d4..c8a943b 100644
--- a/backend/requirements-runtime.txt
+++ b/backend/requirements-runtime.txt
@@ -2,7 +2,7 @@
 # binary), so openai-whisper/torch are intentionally absent — that is the ~2GB
 # the native install saves. --engine whisper-py remains a dev/source-only option
 # via backend/requirements.txt.
-opencv-python-headless>=4.8.1.78
+opencv-python-headless>=5.0.0.93
 numpy>=2.5.1
 # Audio-event detection (YAMNet laughter/reaction channel) runs on ONNX Runtime —
 # no torch/TF, so it stays on the hermetic native path.
diff --git a/backend/requirements.txt b/backend/requirements.txt
index 9fc4013..4715d09 100644
--- a/backend/requirements.txt
+++ b/backend/requirements.txt
@@ -9,7 +9,7 @@ openai-whisper>=20250625
 # speechbrain>=1.0.0
 
 # Face detection (optional, for smart cropping)
-opencv-python-headless>=4.8.1.78
+opencv-python-headless>=5.0.0.93
 numpy>=2.5.1
 
 # Audio-event detection (laughter/reaction channel via YAMNet ONNX — no torch/TF)