AI VIDEO DETECTOR · FREE PLAN
Is This Video Real or AI-Generated?
Face swaps and fully generated clips are getting harder to spot by eye. The AnyFact video detector checks image and sound separately, moment by moment.
Open in beta · 10 free credits a month · Files deleted after 30 days
AI Video & Deepfake Detector
Open in betaWhat you need
Upload a video or paste a link
MP4, MOV or WebM, up to 3 minutes during the beta.
At a glance
Image + sound
Frames, faces and soundtrack each go to their own detector, with suspicious moments marked on a timeline.
AnyFact method
3 levels
Probably human, uncertain or probably AI — never a bare percentage — with the reasons and the limits of the method.
AnyFact method
10 credits
Free every month with an account, no credit card. Uploaded files are deleted after 30 days.
AnyFact free plan
Why your eyes can no longer verify a video alone
A convincing video used to require a studio. Today, face-swap and video-generation tools are widely available, and a short clip shared on a messaging app loses most of the clues that would give it away.
A video is two things at once: images and sound. Either can be generated, and a real image track can carry a cloned voice. Checking them separately matters.
The AnyFact video detector scores every second with the photo detector, faces with an open-source deepfake detector and the soundtrack with the voice detector, then colours suspicious moments on a timeline.
Does this sound familiar?
- A video of a public figure says something out of character.
- A video message asks you to send money urgently.
- A clip is going viral and you want to know before sharing it.
Here's what we check
Upload or paste a link
MP4, MOV or WebM, up to 3 minutes during the beta.
Image and sound, separately
One frame per second is scored; faces and audio get their own detectors.
A verdict in three levels
Probably human, uncertain or probably AI — never a bare percentage — with a short explanation and the limits of the method.
How to read the result
The signals point to a person. It is not a guarantee: a careful generator or heavy editing can leave few traces.
The signals disagree or are too weak. We say so rather than guess, and list what was found.
The signals point to generation. It is an indication, not proof — check the context before acting.
Sample layout. Real reports name the signals that were checked and the limits of the method.
How to detect a fake video, step by step
Upload the video or paste a link to its public page, then read the timeline before the overall verdict. Deepfake video detection works best on the best copy you can find.
- 01Get the file, or the link to the public page where the video was posted.
- 02Upload an MP4, MOV or WebM file up to 200 MB and 3 minutes. For a longer video, trim it to the part that matters.
- 03Read the timeline, where suspicious moments are marked.
- 04Check the face result and the soundtrack result separately.
- 05Trace the source: who posted it first, and when.
A video analysis uses 5 credits.
What does an AI video detector analyse?
It splits the video into the layers that can each be faked, and sends each one to its own detector.
| Layer | How it is checked | Good to know |
|---|---|---|
| Provenance | C2PA content credentials and metadata are read first | Re-uploaded videos usually lose them |
| Frames | One frame per second is scored by the photo detector | Points to fully generated scenes |
| Face | A face-swap classifier checks the largest face | Runs only when a frontal face is clearly visible |
| Soundtrack | The voice detector scores the audio and checks the words for scam patterns | Catches a cloned voice over real footage |
| Second opinion | In the grey zone, Claude reviews the 5 most suspicious frames | It can move the score only a little and never decides alone |
How to read a fake video detector timeline
The shape of the suspicious moments matters as much as their number. Look at where they fall and which layer raised them.
| What you see | Likely reading |
|---|---|
| Flags on the face result while the frames look normal | A possible face swap on real footage |
| Most frames flagged from start to finish | A possibly fully generated clip |
| Frames clean, soundtrack flagged | Real images with a cloned or replaced voice |
| A few isolated seconds flagged | Can be transitions, motion blur or on-screen graphics; look at those moments before concluding |
| No face result | No clear frontal face was visible, so the face check did not run |
Is this video AI? Checks you can do yourself
Pause and look closely at the places where generation and face swaps struggle. These checks work on any video player.
- Watch the mouth during speech: do lips, teeth and tongue match the sounds?
- Pause when the head turns: look at the edges of the face, hairline, glasses and earrings.
- Compare the light on the face with the light in the rest of the scene.
- Read any text in the background and look at hands, which are often distorted.
- Search for the earliest copy of the video and the account that posted it.
Our guide on how to spot a deepfake video shows each check with examples. Fact-checkers combine it with source tracing, as described on our page for journalists and fact-checkers.
When does deepfake video detection struggle?
Results are less clear when the video gives the detectors little to work with. Expect more uncertain results with low-resolution or heavily compressed clips, faces seen in profile or from far away, strong filters, screen recordings of another screen, and very short clips.
Before uploading, look for the best version you can find: the original post rather than a re-share, and the full frame rather than a cropped excerpt. When the layers disagree, give more weight to the one that had the most to work with, such as a clear face or a clean soundtrack.
When only the sound is in doubt, the AI voice detection page explains the soundtrack check in more detail.
Before you check
Is the result proof that a content was made by AI?▼
No. It is an indication, not proof. Detectors can be wrong in both directions, especially on compressed, cropped, edited or very short content. Use the result to decide what to check next, not as a final judgement on anyone.
Why no percentage?▼
A single percentage suggests a precision no detector has. You get one of three levels — probably human, uncertain or probably AI — with the signals behind it and the limits of the method.
How does the analysis work?▼
When the content carries provenance data, such as C2PA content credentials or metadata, we read it first. Then an open-source detector scores the content. When the score is borderline, Claude reviews it as a second opinion, and the result is explained in plain words.
What happens to my files?▼
They are processed to run your analysis — including, for borderline cases, a review by Claude, Anthropic’s model — and deleted after 30 days.
What does it cost?▼
There is a free plan: a free account gets 10 credits every month, with no credit card. Pro ($15 a month) gives 1,000 credits a month; Business ($49 a month) gives unlimited Business Checks and AI checks under a fair-use limit.
Can I paste a link to a video on a social network?▼
Yes, if the link points to a public video page. If the video is private or behind a login, download it where the platform allows and upload the file instead.
Why was the face not checked?▼
The face-swap classifier runs only when a frontal face is clearly visible, and it checks the largest face. Profiles, small or partly hidden faces are skipped. The frames and the soundtrack are still analysed.
Can it check a video longer than 3 minutes?▼
Not in one analysis. Trim the video to the part you want to verify, up to 3 minutes and 200 MB.
Does it detect lip-sync deepfakes?▼
There is no dedicated lip-sync check. A lip-synced video may still show up through the soundtrack, if the voice is cloned, or through frame anomalies, but a clean result does not rule it out. Watching the mouth closely during speech remains a useful manual check.
Test a video free — no account needed for your first analysis
What you need
Upload a video or paste a link
Free plan — 10 credits every month, no credit card.
Each result is an indication, not proof.
Files deleted after 30 days.
You might also want to check
Deepfake Detector: Is This Video, Photo or Voice Real?
Deepfake detection for videos, photos and voice messages: face swaps, generated clips and cloned voices, checked with open-source detectors and explained. Free plan.
Is This Voice Real or AI-Cloned?
AI voice detector for voice notes and recordings: each segment is scored for cloned or synthetic speech, and the words are checked for scam patterns. Free test.
AI Image Detector: Was This Image Made by AI?
Free AI image detector and checker: find out if a photo was generated by AI. Reads C2PA content credentials, runs an open-source detector and explains the verdict.