Fact-check like a newsroom.

The 60-second check

Most claims don't need a deep investigation. They need one honest minute. Before believing or sharing anything consequential, run this sequence:

1. 01 Read past the headline

   Open the actual story. Headlines are written to be shared; bodies are where the hedges, corrections, and context live. If the body doesn't support the headline, you're done - it fails.

2. 02 Check the date

   A remarkable share of "breaking" outrage is a years-old story recirculating. The date is the cheapest fact-check that exists.

3. 03 Identify the original source

   Who reported this first - not who you saw it from? If the trail ends at an anonymous account or a screenshot, treat the claim as unestablished.

4. 04 Search laterally

   Open a new tab and search the claim's key terms. If it's real and significant, independent outlets will have it. If only one ecosystem of sites is carrying it, be suspicious.

Tracing a claim

When the quick check isn't enough - the claim matters and remains unclear - trace it the way we do:

• Find the primary source. If a story cites a study, open the study. If it quotes a statement, find the full statement. Summaries mutate at every hop; the original settles arguments.
• Check the quote in context. Search the exact wording. Clipped quotes routinely reverse meaning - the sentence before or after often contains the qualifier that changes everything.
• Confirm independence. Ten outlets repeating one wire report is one source, not ten. Look for genuinely separate reporting: different bylines, different named sources, different details.
• Check whether it's already been traced. Snopes, Reuters Fact Check, AP Fact Check, and FactCheck.org have often already done the work on the exact claim in your feed.
• Check for the deleted original. If a post or article was edited or removed, the Wayback Machine and archive.today show what it said before.

Checking images

The single most useful verification skill is the reverse image search. It answers the question that catches most fake imagery: has this picture existed before this story?

1. Right-click the image and search with Google Lens, or upload it to TinEye.
2. Sort results by date where possible and find the earliest appearance.
3. Compare the original context to the claimed one. Wrong year, wrong country, or wrong event ends the check.

If the image is genuinely new, interrogate the frame itself:

• Location details. Signage language, license plates, road markings, architecture, and vegetation should match the claimed place.
• Weather and light. Snow in the claimed summer, or long shadows at claimed noon, are quiet contradictions people miss.
• AI artifacts. Garbled text, mangled hands, impossible reflections, and melted background detail still betray many synthetic images - but absence of artifacts proves nothing. Provenance beats pixel-hunting.

Checking video

Video feels like proof, which is why it is the favorite vehicle for recycled and miscaptioned content. The checks are similar to images, with two additions:

• Screenshot key frames and reverse-search those. This catches old footage wearing a new caption - the most common video deception by far.
• Watch the cut points. What happened immediately before the clip starts? Abrupt starts and mid-sentence cuts often hide the context that changes the story.
• Listen. Sirens, language, accents, and background chatter should match the claimed place and moment.
• For dramatic claims, wait for a second angle. Real major events generate footage from many phones within minutes. A singular, unrepeatable clip of a huge event is a warning sign.

Checking accounts

On social platforms, the account is part of the evidence. Before trusting a post:

• Age and history. Was the account created last week? Does its history match its claimed identity - or is it a "local eyewitness" that posted about three other countries this month?
• Prior accuracy. Scroll back. Serial posters of debunked content tell you what their next post is worth.
• Impersonation. Check the handle character by character. Lookalike accounts swap letters, add underscores, and borrow real avatars and names.
• Badges prove little. On platforms where verification is purchased, a checkmark is a receipt, not a credential.

Red flags

Any one of these should slow you down. Two or more together usually decide the question:

• The story makes you furious or triumphant instantly - engineered emotion is a delivery mechanism.
• "Share before this gets taken down" or any other urgency pressure.
• No named sources anywhere: just "reports," "sources say," or "people are saying."
• A screenshot of a headline or post, with no link to the original.
• The domain is one letter off from a real outlet, or a lookalike subdomain.
• Perfect, cinematic footage of a chaotic event, from exactly one angle.
• A statistic with no baseline, no source, and suspiciously round numbers.
• Only one cluster of ideologically aligned sites is carrying a "major" story.

Hard cases

Breaking news

The first hours of any major event follow a pattern: early casualty figures are wrong, eyewitness counts conflict, officials revise, and recycled footage floods in. This is normal, not conspiracy. Treat every early detail as provisional, favor primary feeds (USGS, weather services, official emergency channels), and expect the record to move. It is exactly why our updates are labeled developing or confirmed, never both.

Health and science claims

One study proves little; science moves by accumulation. Check whether the study involved humans or mice, how large it was, who funded it, and whether the headline claims more than the paper does - it usually does. "Peer-reviewed" is the floor, not the ceiling.

Political claims

Assume every viral political claim is optimized against you checking it. Find the full quote with its question, the full vote with its context, the full bill rather than a summary of a summary. Both flattering and damning versions of the same event usually omit something - find what.

The checklist

Before sharing anything consequential, you should be able to answer yes to each of these:

• I read the actual story, not just the headline.
• I know the date, and the story is actually current.
• I know who reported it first, and they are in a position to know.
• At least one genuinely independent source confirms the core claim.
• Images or video check out - reverse-searched, right event, right time.
• Quotes and numbers survive contact with their original context.
• I would share this even if it cut against my own side.

That last one is the honest tell. Misinformation spreads because it flatters someone's priors - and on the stories you most want to be true, you are the easiest person to fool.