AI Video Detector logo AI Video Detector

Evidence Lab for Social Video

Check if a video may be AI-generated

Paste a public link or upload a video. Review the evidence, confidence, and what could not be checked before you share it.

Estimates, not proof. Some social links cannot be accessed. Uploading the original video works best.

AI video check

Start with a link or upload a video

Link access is best-effort. Original uploads receive the strongest evidence scan.

Fastest

Best-effort for supported public social video URLs. If public media is unavailable, upload the original video for deeper analysis.

By clicking Start free check, you confirm you have the right to submit this video link for analysis and understand results are likelihood-based, not proof.

Estimated credits depend on video size and duration. Detected platform: Unknown.

How the check works

Live analysis process

Follow your video check

See what is happening at each stage. Scores appear only for areas that can be checked.

Likelihood-based · not proof
  1. 1Input
  2. 2Access
  3. 3Media
  4. 4Queue
  5. 5Analyze
  6. 6Report

Check progress

Step 0 of 6

How the check works

Paste a public video link or upload a video you have the right to analyze. We show what can be checked, what may be limited, and how to read the result.

Ready when you are · stage-based progress

Check progress

What is happening now

Each stage shows its current status. Timing depends on video access and plan limits.

Stage-based
  1. 1

    Checking input

    Validate the URL, file, consent, and supported video source.

    Pending

    0% · Pending

  2. 2

    Checking access

    Run security checks and see what media can be reached safely.

    Pending

    0% · Pending

  3. 3

    Preparing video

    Get the available video ready for review.

    Pending

    0% · Pending

  4. 4

    Waiting to start

    Wait for the check to begin. Paid plans start sooner during busy periods.

    Pending

    0% · Pending

  5. 5

    Reviewing evidence

    Review picture quality, movement, source details, and sound when available.

    Pending

    0% · Pending

  6. 6

    Building report

    Assemble likelihood, confidence, limitations, and recommendations.

    Pending

    0% · Pending

Signal checks

Evidence signal readiness

While the check runs, this panel shows progress only. The completed result appears below.

Process status

Source/media access

Waiting for the analyzer to confirm source and media access.

Pending

Pending

Visual clues

A score appears after enough of the picture can be reviewed.

Pending

Pending

Movement consistency

A score appears after changes over time can be reviewed.

Pending

Pending

Picture quality clues

A score appears after the video quality can be reviewed.

Pending

Pending

Sound and picture consistency

A score appears when both sound and video are available.

Pending

Pending

Source details

Available information about the video and its source will appear here.

Pending

Pending

Why it helps

Check suspicious videos before you share them

Some videos look real at first glance. Then something feels off: the face moves strangely, the hands blur, the audio feels inconsistent with the clip, or the clip appears during a breaking news event with no clear source. AI Video Detector gives you a quick way to slow down.

No manual download when a public link works

Paste a social video link first. If the link cannot be checked, you will see a clear reason and a safer next step.

See evidence before deciding

Start with a free preview. Use credits to unlock the available explanations and PDF of your unexpired report without uploading again.

Built for cautious sharing

The result helps you decide whether to pause, look for the original source, compare other uploads, or ask for human review.

Clear failure states

Private, deleted, restricted, or unavailable videos are not silently treated as real or fake. You get a plain explanation.

Supported checks

Choose the right AI video check for your input

Start with the page that matches where the suspicious video came from. Public-link checks are best-effort, and upload is the stronger option when you have the original file and the right to analyze it.

YouTube AI Video Detector

Check public YouTube videos or Shorts for available link and video signals. Some YouTube links may return limited public-link signals, so upload the original file when you need deeper frame review.

Check YouTube videos →

TikTok AI Video Detector

Try a public TikTok video link when the media is accessible and within scan limits. Private, removed, login-only, or region-limited videos may fail safely or ask for upload.

Check TikTok videos →

Instagram AI Video Detector

Review public Instagram Reels or video posts on a best-effort basis. Stories, private accounts, photo-only posts, and login-only content are not reliable inputs.

Check Instagram Reels →

X / Twitter AI Video Detector

Check public X or Twitter posts that contain video. Protected, deleted, login-only, blocked, or no-video posts should return a clear limitation instead of a misleading score.

Check X / Twitter videos →

Deepfake Video Detector

Focus on deepfake and AI-manipulated video signals, including visual, motion, audio-video, source, and limitation clues. Results are likelihood-based, not proof.

Check deepfake video signals →

Detect AI Video From a Link

Use the generic link workflow when you want to paste one public social video URL first and understand whether link-based scanning is possible.

Detect AI video from a link →

Original Video Upload

Upload the original video file when a public link is limited or when you need the strongest available frame, motion, duration, and audio review. Only upload files you have the right to analyze.

Upload original video →

Evidence report

What your evidence report shows

An evidence report explains the signals behind the likelihood score, not just a yes-or-no label. It helps you see what looked suspicious, how confident the scan is, and what the scanner could not check.

First check the media access status and frames analyzed. A report can include likelihood, confidence, sampled visual and motion signals, audio checks when available, limitations, and next steps. A limited report without readable video has no AI score.

A high score is not legal proof. A low score does not prove a video is real. Use the report as a first-pass screening aid and review the evidence yourself.

Learn how reports workCheck deepfake video signalsCompare report depth

Reproducible sample · same video, same analysis

See what a report unlock adds

Watch real camera footage of a duck swimming in a pond, then compare the free preview with the full report from the same analysis. This licensed public sample is not a customer upload or a detection accuracy benchmark.

Video: Duck swimming at Knollenteich by PantheraLeo1359531 · CC BY 4.0. Resized and transcoded; source color-corrected by the author. Source attribution does not guarantee authenticity or imply endorsement.

3 sampled frames at 1.549, 4.646, 7.743 seconds · modal-ffmpeg-cv-audio-v11 · basic audio check

Free preview

Screening index: 44/100 · Inconclusive

Confidence indicator: 78/100. Neither value is a calibrated probability of AI generation.

  • Metadata reliability

    ffprobe extracted container, stream, codec, duration, and bitrate metadata.

    Signal value: 0.75

  • Compression anomaly

    Frame sharpness and histogram changes are checked for unusual compression/artifact patterns.

    Signal value: 0.078

4 additional recorded explanations: Temporal consistency, Texture entropy, Edge consistency, Audio-video consistency.

All limitations · visible in both
  • Visual indicators are rule-based screening signals, not a calibrated probability that a video is AI-generated.
  • A few sampled frames cannot cover every moment; normal scene changes can affect visual indicators.
  • This is a signal-based screening report, not proof.
  • CPU ffmpeg/OpenCV signals are weaker than a dedicated forensic or GPU deepfake model.
  • Social page URLs that cannot be safely fetched are not bypassed; upload/direct video URLs produce stronger reports.

Full · same existing report + PDF

Screening index: 44/100 · Inconclusive

Confidence indicator: 78/100. Neither value is a calibrated probability of AI generation.

  • Metadata reliability

    ffprobe extracted container, stream, codec, duration, and bitrate metadata.

    Signal value: 0.75

  • Compression anomaly

    Frame sharpness and histogram changes are checked for unusual compression/artifact patterns.

    Signal value: 0.078

  • Temporal consistency

    Sampled frames are compared for frame-to-frame consistency.

    Signal value: 0.775

  • Texture entropy

    Sampled frame luminance entropy is used as a lightweight texture signal.

    Signal value: 0.688

  • Edge consistency

    Canny edge density variation is used as a simple artifact proxy.

    Signal value: 0.966

  • Audio-video consistency

    Audio presence, duration alignment, silence, clipping, and level stability were checked with a low-cost audio sample. Paid reports can add sampled audio-motion windows. Semantic speech/video matching is not performed.

    Signal value: 0.003

Unlocking reveals these existing explanations and PDF export. It does not rerun analysis, add Pro audio analysis, or extend expiry.

All limitations · visible in both
  • Visual indicators are rule-based screening signals, not a calibrated probability that a video is AI-generated.
  • A few sampled frames cannot cover every moment; normal scene changes can affect visual indicators.
  • This is a signal-based screening report, not proof.
  • CPU ffmpeg/OpenCV signals are weaker than a dedicated forensic or GPU deepfake model.
  • Social page URLs that cannot be safely fetched are not bypassed; upload/direct video URLs produce stronger reports.
Inspect the actual sample output and reproducibility hashes

Workflow

How it works

1

Paste a public link or upload a video

Start with a public link when available. For the strongest report, upload the original video file.

2

Review what can be checked

If the link is private, deleted, restricted, or unavailable, we explain the problem instead of showing a misleading score.

3

Read the evidence report

See likelihood, confidence, suspicious signals, unchecked signals, and safer next steps.

Pricing

Pay once or choose Pro

Start with a free daily check, buy credits without a subscription, or choose monthly or annual Pro for regular use.

Pay once · start here

Unlock complete reports without a subscription

Start with a free result. Buy just that eligible report, or choose credits for occasional checks. Failed checks do not consume credits.

No recurring charge

Free

$0

Check an accessible video first. See the screening result, confidence, sampled coverage, two signal previews, and all limitations.

  • Daily limits and availability apply
  • No card required
  • Restricted links may need an upload
Start free check

Single report

$2.99USD · once

Unlock the remaining evidence and PDF for one eligible, already-generated report. Available from your result when there is useful evidence to unlock.

  • Bound to that report and your account
  • Existing report expiry still applies
  • No new analysis, credits, or Pro upgrade
  • No subscription
Check a video first

Starter Pack

$9

20 credits · pay once

For occasional checks. Use credits for new video analysis or an eligible existing report. Review the estimated cost before confirming.

  • Up to 250 MB / 3 minutes per video
  • Credits valid for 12 months
  • Credits depend on size and duration
  • 20 credits does not mean 20 videos
  • Failed checks do not consume credits
Check videos regularly? See Pro and 100-credit top-up

Pro includes 100 monthly credits, larger limits, priority processing, longer report history, and deeper sound and movement review. Your existing credits and purchased reports remain available under their original terms.

Pro Monthly

Flexible

For creators, journalists, and teams that check videos regularly.

$29.99/month

Billed monthly

Monthly credits100 credits
Maximum file sizeUp to 500 MB
Maximum video lengthUp to 5 minutes
ProcessingFaster during busy periods
Report history90 days
Report detailDeeper sound and movement review
PDF downloadsIncluded

Pro Annual

Save $119.88/year

For creators, journalists, and teams that check videos regularly.

$240/year

Equivalent to $20/month · billed annually

Monthly credits100 credits
Maximum file sizeUp to 500 MB
Maximum video lengthUp to 5 minutes
ProcessingFaster during busy periods
Report history90 days
Report detailDeeper sound and movement review
PDF downloadsIncluded

100-credit top-up · $29 once

Valid for 12 months. Video size and duration determine credits per check.

Credits are based on video size and duration.

One-time credits unlock complete successful reports. Pro adds recurring capacity, priority, longer retention, and deeper report features.

Free checks are subject to daily limits, queue availability, and abuse controls.

Link checks depend on public media access; upload is the most reliable path for full video scans.

AI video detection is a screening workflow, not legal or forensic proof.

FAQ

Common questions

What is an AI video detector?

An AI video detector is a tool that checks whether a video may show signs of AI generation, deepfake editing, or other manipulation. AI Video Detector supports public social link checks with platform limits and original video uploads for deeper review.

How do I check if a video is AI-generated?

Try a public social video link or upload the original video file, run the check, and review the evidence report. Link scans depend on public media access; uploads provide the strongest available review.

Can I detect AI video from a link?

Yes, for supported public links when the video can be safely accessed. Some links return limited public-link signals only; upload the original video for deeper analysis.

Is AI Video Detector a deepfake video detector?

Yes, it can check for deepfake-related and AI-manipulated video signals. It also covers broader AI-generated video signals, source/media limits, confidence, and unchecked signals. The result is likelihood-based and not proof.

What checks are supported?

AI Video Detector supports public social video link checks for YouTube, TikTok, Instagram, and X / Twitter on a best-effort basis, plus original video uploads when you have rights to analyze the file. Some public links may be limited or unavailable.

Can AI video detectors be wrong?

Yes. Compression, edits, reposts, low quality, short clips, missing audio, and partial AI editing can all affect results. Treat the report as a screening aid.

Can this prove a video is fake?

No. The report does not prove a video is fake or real. It shows signals and confidence so you can decide whether more review is needed.

Does AI Video Detector check audio?

Yes, when usable sound is available. Free reports include a basic sound review, while Pro checks can provide deeper sampled sound and movement comparisons. A single-report unlock reveals existing analysis and does not add Pro analysis. The report clearly marks anything that could not be checked.

Why was audio-video consistency not checked?

Some videos have no readable audio, muted audio, restricted media access, or platform-transcoded audio that cannot be sampled reliably. When audio is unavailable, the report marks the signal as not checked and does not treat it as low risk.

Evidence notes

Practical guides for suspicious social videos

Read the context, preserve the source, understand the limits of a detector, and know when the claim needs more than an automated result.

Browse all guides