Everyday Telling good information from bad
Checking a viral claim before you pass it on
The five minute routine for a shocking post: find the origin, check the date, verify the image, and read what is missing.
The short answer
- Five minutes on four questions catches most viral falsehoods: where it first appeared, when it happened, whether the picture belongs to the story, and what is being left out.
- The most common fake is not fake at all: it is a real photo or clip with a caption that moves it to another country, cause or year.
- Reverse image search is the highest value single step, because a picture that existed before the event it supposedly shows settles the matter instantly.
- A screenshot proves nothing on its own, since anyone can edit text in a browser and photograph the result in seconds.
- Wording designed to make you feel urgent, vindicated or afraid is a reason to check harder, not a reason to share faster.
- If you already shared something false, correct it in the same place you posted it rather than deleting it quietly.
Before you forward it, spend five minutes on four questions: where did this first appear, when did it actually happen, is the picture really from this event, and what is being left out? Most viral claims fall apart on one of those four, usually the second. You do not need special tools or any expertise in the subject. You need a search box, a reverse image search, and a willingness to wait long enough to find the version of the story that arrived before the outrage did.
The five minute sequence
Do these in order. Stop as soon as one of them answers the question.
- Describe the claim in your own words and search that. Not the hashtag, not the caption. Plain words: the place, the person, the thing that supposedly happened. If something this big occurred, other people are writing about it in ordinary language.
- Find the earliest version you can. Sort results by date, or add a rough time window. You are looking for the first telling, not the loudest one.
- Reverse image search any picture. Drag the image into an image search tool, or use your phone's built in lens feature. You are asking one question: has this picture been online before today?
- Check the date and the place. Weather, foliage, language on shop signs, license plates, the style of police uniforms, whether people are wearing coats. These beat captions, because a caption is just text somebody typed.
- Reach a primary source. The court filing, the company statement, the official account, the actual video. One document read directly is worth twenty summaries, and a generated summary only counts once you open its citations and check them.
- Ask what is missing. The sentence before the quote. The denominator under the number. The year. The other party's response.
If the claim survives all six, it is probably worth repeating, with a note about where you got it. If you ran out of patience at step two, that is a result too: unverified, not shared.
Find the earliest version, not the loudest
The post you saw is rarely the original. Viral material travels by copying: somebody screenshots, recaps, reposts with a hotter caption, and the details drift at every step. By the time it reaches you the claim has often changed in a specific and predictable way, growing more extreme and losing its qualifiers.
Working backwards is the single highest value habit here. Search a distinctive phrase from the caption in quotation marks and sort by oldest available. Look for the account that first posted the media rather than the account with the most reshares. When you find the origin, two things usually become clear at once: what actually happened, and which step in the chain added the part that made you angry.
Watch for the case where the earliest version is itself a joke. Satire accounts get screenshotted without their labels, and the next hop strips whatever made it obviously a joke. A parody handle or a satirical bio on the original is your answer.
Checking a photo or a video
Reverse image search is the workhorse. Save the picture, or take a screenshot of a video frame, and search it. Some tools let you sort matches by oldest, which is exactly what you want: if a photo presented as yesterday's protest appears on a news site from four years ago, you are finished.
Metadata is less useful than films suggest. Social platforms strip the embedded camera data from uploads, so the absence of it proves nothing. Some images now carry signed provenance information instead, and a Content Credentials label showing where a picture came from is worth checking when it is present, though most images still have none.
For fully synthetic media, judge the whole frame rather than hunting for a single flaw. Lighting that comes from two directions, text on signs that dissolves at the edges, jewelry and teeth that change between frames, and backgrounds that repeat are the usual giveaways, and there is more detail in how to tell whether an image was generated. Video has its own tells, mostly around physics and continuity: objects that pass through each other, crowds whose members morph, and clips that are suspiciously short because length is where generated AI video still falls apart. Audio is the hardest of the three, since a convincing voice clone needs only a few seconds of source audio, so a voice note alone should never settle anything.
Real media with the wrong story attached
The most common fake is not fake at all. It is a genuine photograph or clip, unaltered, with a caption that moves it to a different country, a different year or a different cause. This works because the image survives every authenticity check you can run. There is nothing to detect. The lie lives in the caption.
Recontextualized media shows up after every disaster, conflict and election, often recycling footage from an unrelated event elsewhere in the world. The counter is always the same: search the picture, not the story. If the earliest appearance predates the event being described, the caption is wrong however real the image is.
| What you are looking at | Fastest check | What it commonly turns out to be |
|---|---|---|
| A shocking photo of a disaster | Reverse image search, sorted by oldest | A real photo from a different event or an earlier year |
| A dramatic video clip | Search the event in plain words and look for a longer cut | Real footage cropped to remove what happened just before |
| A screenshot of a post or headline | Search the exact words and look for the live original | A real post edited in a browser, or one deleted hours later |
| A striking statistic | Search the number with the body said to have produced it | A real figure about a different country, year or population |
| A quote on a photo of a public figure | Search the quote in quotation marks | Misattributed, often to a person who never said it |
| Breaking news from an account you do not know | Open the profile and read its last twenty posts | An engagement account recycling old news for reach |
Screenshots, deleted posts and missing context
A screenshot is the weakest form of evidence in circulation, because anyone can edit the text of a live web page in their browser and photograph the result in about fifteen seconds. Treat every screenshot as unverified until you find the thing it claims to show.
Sometimes the original really was deleted, which does not automatically mean it was real. Search the exact wording plus the author's name and see whether anyone archived it, or whether reporters quoted it while it was up. Genuine deleted posts from public figures leave a wide trail, and invented ones leave a single screenshot with no witnesses.
Missing context is quieter and more common. A number without a denominator, a percentage change without the starting point, a quote that answers a question you were not shown, a rule described without the exemption that follows it. When a claim seems both true and outrageous, the useful question is not whether it happened but what came immediately before it.
Emotional framing is the tell
The strongest single predictor of a claim being distorted is not its subject. It is how hard the wording works on your feelings. Phrases like nobody is talking about this, share before it is taken down, and they do not want you to see this are not incidental. They are instructions to skip verification.
There is a version of this you can use on yourself. If a post makes you feel immediately vindicated about a group you already dislike, raise your checking standard rather than lowering it. Material aimed at your own side gets shared with the least scrutiny, which is exactly why it is made.
The same pressure appears in fraud, because it works on the same wiring. Urgency, secrecy and a closing window are the shared grammar of viral misinformation and scams built with AI generated voices and faces, and recognizing the pattern in one place makes it easier to spot in the other.
What to do if you already shared it
Correct it visibly, and do it where you posted it. Deleting quietly is the worst option, because the copies other people made survive and nobody who saw it learns anything. Post a short correction in the same place, in plain words: this was wrong, here is what actually happened, here is the source. No defensiveness and no long explanation of how you were misled.
In a group chat, reply directly under the original message so the correction sits with it. If you sent it to one person, tell that person yourself rather than hoping they missed it. Corrections spread less far than the original claim, which is a reason to make them promptly rather than a reason to skip them.
Then make the next one cheaper. A bookmark folder holding a reverse image search tool and a couple of fact checking organizations turns the routine above into a two minute job. If you keep landing on unfamiliar websites, the broader skill of working out whether a source is worth believing at all saves more time than checking claims one at a time.
Common questions
How do I reverse image search on a phone?
Save the picture to your photo library first, or screenshot a frame from the video. Then open the lens or camera feature inside your search app or browser and choose the image from your library. Most phone browsers also let you press and hold an image on a web page and pick a search option directly from the menu that appears.
Is it enough if a major news outlet reported it?
It helps a lot, but check whether the outlet did its own work or simply repeated a social media post. Look for a named reporter, a dateline, and quotes from people the outlet spoke to. Early coverage of a fast moving event is frequently corrected later, so a report published in the first hour deserves less weight than the same outlet's account a day afterwards.
What if I cannot find the original post anywhere?
Then the claim stays unverified, which is a legitimate place to leave it. Absence of evidence is not proof of invention, but it does mean you have nothing to pass on. Say so if you mention it at all. Real events of any size normally leave multiple independent traces within hours, so a total absence of them is meaningful.
Are fact checking websites biased?
Check their method rather than their reputation. A good check names its sources, links to primary documents, and shows the reasoning so you can disagree with the conclusion while still using the evidence. Use the documents they surface rather than their verdict. Where two organizations disagree, the difference is usually about how a claim was worded, not about the underlying facts.
Does sharing something to debunk it still spread it?
It can, so lead with the truth rather than the falsehood. Put the accurate version in your first sentence, keep the false claim short, and avoid reposting the original image if you can describe it instead. Repeating a false statement prominently, even to deny it, tends to make it more familiar and familiarity feels like truth.