Prefer the original video file
Very short clips provide too little information. Prefer a complete, playable source video that has not been transcoded repeatedly.
Read the human verification guideVideo frames, metadata, and AI verification Text · Image · Video
AI video detector: check frames and generation clues
Choose a video up to 120 seconds and 100MB. After upload consent, the service checks up to six sampled frames and reports coverage limits; audio and temporal continuity are not analyzed.
- Three content types
- No login or sign-up
- Evidence you can inspect
Content screening and verification
Choose the content to inspect
Results separate heuristic screening signals, file facts, and evidence limitations.
Text content
Paste the text you want to inspect
GenDetect checks repetition, sentence length, punctuation, and character structure.
Your text is not saved or uploaded
Image content
Choose a local image
Automatically checks generator metadata, provenance credentials and model signals, with consent before server analysis.
本地启发式不会上传图片。
Server analysis requires confirmation of the upload and processing notice
Video content
Choose a local video
Server analysis supports videos up to 120 seconds and 100MB, inspecting at most six sampled frames after upload consent. Audio and temporal continuity are not checked.
Analysis time depends on file size
Video-frame and metadata guide
Can sampled frames identify AI-generated video?
Detection follows enabled capabilities. Server analysis requires upload consent and checks up to six sampled frames. Provenance, model signals and coverage limits are reported separately; continuity and audio need separate review.
Start with media and file details
Server analysis checks up to six frames, records timestamps and displays independent scores. Uncalibrated results remain inconclusive. Local checks apply only when server detection is not enabled.
Read methodology and limitationsVerify continuity and source
Also review motion, lighting continuity, audio synchronization, the publishing account, and the original file before reaching a conclusion.
Review accuracy and false positivesDetection methods
What signals does AI content detection examine?
Each content type needs a different analysis. GenDetect separates screening signals, file facts, and limitations.
AI text detection
Examines vocabulary repetition, recurring phrases, sentence-length distribution, punctuation frequency, and unusual characters. More complete text usually produces steadier statistical signals.
Open text detectorAI image detection
Checks original-file metadata, provenance credentials and model signals. Unverified claims and uncalibrated scores remain supporting clues.
Open image detectorVideo-frame and metadata screening
Server analysis checks at most six frames of videos up to 120 seconds. It does not inspect audio or temporal continuity, and a negative result does not establish human origin.
Open video detectorHow to read the score
The screening index is a rule anomaly score, not a source probability.
This demo locally aggregates heuristics for text, images, and sampled video frames; images also check known generator metadata. The index is not trained or probability-calibrated and cannot prove origin.
Frequently asked questions
Video metadata and AI verification FAQ
Learn the scope of free metadata screening and why deepfakes or a specific generator such as Sora cannot be identified from file clues alone.
What does the current video detector inspect?
Local mode samples around 20%, 50%, and 80% of the video, preferring nearby keyframes when supported and analyzing duplicate stills only once, for at most three static frames. It also checks content credentials and reports file size, duration, resolution, estimated bitrate, MIME type, and extension. Optional full detection, when enabled and with upload consent, analyzes up to six sampled frames from videos up to 120 seconds long. Neither mode treats rule signals as AI probabilities.
Is this a deepfake video detector that inspects video frame by frame?
No. Local mode checks at most three static frames; optional full detection analyzes up to six sampled frames. Neither mode performs frame-by-frame, motion-continuity, or audio semantic analysis, so it cannot independently identify face swaps, lip-sync errors, synthetic speech, or other deepfakes.
Can it tell whether Sora or a specific AI model generated a video?
Not reliably. Different generators, camera footage, editing, and transcoding can produce similar sampled-frame statistics, so the tool cannot attribute a video to Sora or another specific model. Verify temporal continuity, the original file, content credentials, and publishing source.