Sentinel.AI

Deepfake detector for images, voice and video

Connecting... 🎥 Live camera

📂

Drop a file here, or tap to choose

Image, voice recording or video with sound. Paste an image with Ctrl+V.

🖼️ JPG · PNG · WEBP🎙️ WAV · MP3 · M4A🎬 MP4 · MOV · WEBM

Model

Options Crop to face first

Try this for wide shots where the face is small.

1Upload

2Analyse

3Result

Uploading...

REALdecision boundaryFAKE

Fake probability

Confidence

Analysed as

Model

Audio used

Time

Technical details

For the best results

Upload the original file. Re-sent or screen-recorded media loses the traces the model looks for.
Keep clips short (3 to 10 seconds) with a clear face and, for video, speech.
Results are probabilities, not proof. If the answer is Uncertain, get a human check.
If a result looks wrong, try the other model or the face-crop option.