Community Notes and AI videos on X

A Community Note under a video on X can be the quickest sign that a clip is AI-made, old or misleading. But notes take time to appear, most posts never get one, and a note is about the claim rather than a check of the footage itself. Here's how to read them, and what to do when there isn't one.

How Community Notes work

Notes are written by volunteer contributors, not by X staff. Other contributors then rate each note. A note is shown publicly only when enough contributors rate it helpful, and, importantly, when those ratings come from people who usually disagree with each other. The idea is that a note people on different sides both find helpful is less likely to be one-sided.

A few details matter for videos. These are based on X's own announcements, and features change, so treat them as a guide rather than a manual:

• A note can attach to the video itself. Contributors can mark a note as being about the media, so it can appear on other posts that share the same video, not only the post where it was written.
• You can ask for a note. At the time of writing, X lets users with a verified phone number request a note from a post's menu. Contributors then decide whether to write one.
• Notes can change. A note shown today can disappear if later ratings push it down.

What a good note tells you

The most useful notes on videos do three things:

1. Say what's wrong in a sentence or two, such as "This video was made with AI" or "This was filmed in another country in an earlier year".
2. Link to sources: the original creator's post, a news report, a fact-check, or the first upload with an AI tool's watermark still on it.
3. Explain the gap between what the post claims and what the evidence shows.

When a note links to the original creator saying "I made this with AI", that's strong evidence, often stronger than any detector's score.

What a note can't tell you

• No note doesn't mean real. Most posts never get one. The rating process needs time and agreement, and a viral clip can collect huge numbers of views before any note is shown.
• Many notes are about context. Plenty of notes on videos deal with where and when a clip was filmed, not whether it was AI-generated.
• Notes can be wrong or incomplete. Contributors are volunteers. Open the sources rather than trusting the summary.
• A note may not follow every copy. Crops, re-edits and screen recordings might not match the original video closely enough for its note to appear on them.

Below: firefighters in breathing gear in front of a fire, made with Alibaba Wan 2.1, from our held-out test set. Our video model scored it 99% AI. Dramatic emergency scenes like this are the kind of clip that can spread fast during breaking news, often before any note has had time to appear.

Video VerifierAlibaba Wan 2.1

Video is likely

AI-made

Test set · Oct 2026crackly-ai.comVideo model: 99% AI

Using a note and a scan together

A note and a scan answer different questions. A note asks whether the claim is true. A scan asks whether the footage looks camera-made. Use both:

1. Copy the post's link from X's share menu.
2. Paste it into the checker. With a link, Video Verifier fetches the video, reads the post's text and tags for AI labels and looks up published fact-checks, alongside its checks of the picture, the motion and the sound. Our X page has more on checking posts from X.
3. Compare the two. If the note says "AI-generated" and the scan says "Most likely AI-generated" with evidence, you have two independent signals. If the scan says "Too close to call", lean on the note's sources and the original creator instead.
4. If they disagree, dig deeper. Don't simply pick the answer you prefer. Read the note's sources, look at which checks drove the scan's result, and find the earliest upload.

When there's no note yet

You may simply be early. Do what contributors do:

• Find the first upload. Search key words from the caption, or search screenshots with Google Lens.
• Read the replies with care. Quote posts and replies sometimes point to the original creator, but they're often wrong too.
• Check the account. Does it post only viral clips with no sources? Has its name or topic changed recently?
• Scan the post and read the evidence, not just the headline.

For breaking-news clips about disasters and conflicts, our guide on fake disaster and war footage has more steps.

If you write notes yourself

A few habits make notes about AI videos stronger:

• Lead with the original creator's post or a visible AI watermark on the earliest copy, where you can find them.
• If you mention a detector's result, describe what it found rather than quoting a bare percentage. Detector results are estimates, and clear evidence is easier for other contributors to judge.
• Say when a clip is real footage with AI edits rather than fully AI-made. The difference matters to readers.

Check the post yourself

Got a video from X in front of you? Copy its link and paste it into our AI or real checker. A Quick Scan is free and answers in under a minute for most videos. For clips made with the tool in the example above, see our Wan page.

Quick answers

Does a Community Note mean the video is fake?

Not always. A note adds context. Some notes say a video is real but wrongly captioned. Read the note and its sources.

Can I ask for a note on a video?

At the time of writing, X lets eligible users request one from the post's menu. Contributors decide whether to write it, and it only shows once it's rated helpful.

Can a scan replace a Community Note?

No, and a note can't replace a scan. A scan looks at the footage; a note looks at the claim. Together they give you a much clearer picture.