Wan AI detector: was this video made with Alibaba's Wan?
Wan, from Alibaba, is one of the most widely used open AI video models: anyone can run it on their own computer, with no watermark and no limits. That makes it popular for clips meant to fool people. Paste a link or upload the clip to check it.
How well we catch Wan
- of 628 Wan test videos flagged as AI
- 96.0%
- called real by mistake
- 1.0%
- each clip tested as published, WhatsApp-quality, 360p and heavily compressed
- 4×
Our model flagged 96.0% of the Wan clips in our main test (Wanx and Wan 2.1) and called 1.0% of them real. In a separate independent test it flagged 98.2% of Wan 2.2 image-to-video clips. Every AI tool's results (measured 2026-10-10).
Signs of a Wan video you can check yourself
- Usually no watermark: the open model run on someone's own computer adds none. Clips from Alibaba's own apps may carry a “通义万相” mark or an “AI生成” label.
- Clips about 5 seconds long that bring a photo to life: the first frame looks perfect, then details drift as things move.
- Faces that slowly change from one person to someone similar, and fingers or teeth that blur in motion.
- Text and signs that are almost readable but not quite.
None of these proves anything on its own: real videos have odd moments too, and the best AI clips have none.
Questions
- Can it detect Wan 2.2?
- In our independent test of October 2026 it flagged 98.2% of Wan 2.2 clips, each checked as published and as a WhatsApp-quality copy. No detector is 100% sure, so every result shows its reasons.
- Why don't Wan videos have a watermark?
- Wan is an open model: people download it and run it themselves, so nothing forces a watermark. That's why the check looks at the video itself rather than relying on labels.
Other AI tools: Sora detector · Google Veo detector · Kling detector · Runway detector · Hailuo detector · Seedance detector · Pika detector · Luma detector · PixVerse detector · HunyuanVideo detector