Verify before you share: a practical guide to checking claims online

A companion article to the course Verify Before You Share: A Media-Literacy Story

Most of us are not professional journalists. We are colleagues, friends, and family members who pass on links because they look useful, urgent, or interesting. That habit is now part of how information moves at work and at home. The problem is not that people are careless. It is that speed is rewarded, and verification takes a pause.

This article sets out a repeatable way to check a claim before you share it. It is written for working adults who already use news sites, messaging apps, and workplace channels, but have had little formal media-literacy training. The accompanying course follows Maya Chen, a communications coordinator racing a 4 p.m. standup, as a politically neutral rumour — a warehouse fire blamed on a new supplier — hits her team channel. The skills below are the same ones she practises.

Learning outcomes

After reading this article, you should be able to:

• classify a statement as fact, opinion, interpretation, or persuasion before you treat it as news
• distinguish misinformation from disinformation without assuming motive too early
• audit an online source using five practical checks
• use lateral reading instead of staying on the first page
• check images, video, and AI-generated media, including the limits of detector tools
• run a five-step verification workflow and assign a confidence level before you post

Why this matters at work

One forwarded link can become the unofficial record of an incident. In Maya’s story, a dramatic post arrives with a cropped photo and an unnamed “internal source.” If she forwards it to look helpful before standup, the rumour can outrun the correction. If she pauses, names the claim, and posts only what she can support, leadership still gets an update — just a bounded one.

The same pattern appears in family chats and team channels every week. The skill is not “never share.” It is “share what you can stand behind.”

What media literacy is and what it is not

Media literacy is the ability to access, analyse, evaluate, and create media, then decide what to do with it. Two nearby ideas are easy to mix up:

• Digital literacy is mainly about using tools: accounts, apps, search, privacy settings.
• Information literacy is mainly about finding, evaluating, and using information for a purpose.

You can be fluent with a phone and still treat a persuasive caption as a confirmed fact. At work, that distinction matters. Being able to open a link is not the same as being able to judge it.

Name the statement type first

Before you fact-check, name what you are looking at:

A mixed post is common: one checkable sentence wrapped in emotional framing. If you treat the whole package as fact, you inherit the persuasion as well.

Activity 1 (5 minutes)
Open the last link you almost shared. Write one sentence that captures the checkable claim, stripping out adjectives and calls to action. If you cannot write that sentence, you do not yet have something to verify.

Misinformation is not the same as disinformation

Researchers and practitioners often use Claire Wardle’s information disorder framing:

• Misinformation is false or misleading information shared without intent to harm. People forward it because it fits a fear, a loyalty, or a deadline.
• Disinformation is false or misleading information shared with intent to deceive.
• Malinformation is genuine information used to cause harm — for example, leaking a private detail to damage someone.

In the first minutes after a claim appears, you usually cannot see motive. You can see spread. The practical move is to slow the claim down first, and infer intent later if evidence appears. Accusing a colleague of lying, when they may only have been fast, often makes the next correction harder.

Why inaccurate information travels

Several well-documented forces help false or shaky claims move:

• Novelty. Unusual stories attract attention.
• Emotion. Outrage, fear, and triumph are more shareable than a cautious update.
• Social identity. We are more likely to pass on something that flatters “our side” of a workplace or community.
• Speed norms. In chats, a fast reply can look more loyal than a careful one.

A 2018 study by Vosoughi, Roy, and Aral in Science found that false news on Twitter spread farther, faster, and more broadly than true news, in part because of novelty and emotional response. You do not need to memorise the paper. You do need the implication: your pause is a public service.

Activity 2
Think of a time you forwarded something and later saw a quieter correction. What made the first post feel urgent? Was it the photo, the headline, a colleague’s name, or the clock?

Five checks for an online source

When Maya finds an article that appears to confirm the warehouse fire, the temptation is to cite it internally because it is convenient. A short audit is faster than a walk-back.

Use these five checks, which sit in the same family as public news-literacy checklists (for example, work associated with the News Literacy Project and similar classroom and civic resources):

1. Author and organisation. Who wrote this, and who publishes them? Can you find a real outlet, masthead, or accountable author — not only a display name?
2. Evidence and citations. Does the piece show documents, data, named officials, or on-the-record sources? “An internal source” is a flag, not a footnote.
3. Date and context. Is this from today, last year, or another incident entirely? Old reporting recirculates in new wrapping.
4. Corroboration. Do independent outlets with different incentives report the same core facts?
5. Emotional manipulation cues. ALL CAPS, miracle certainty, enemy language, and “share before they delete this” are design choices. They are not proof.

You are not awarding a purity score. You are deciding trust, caution, or rejection for this claim, today. Download this checklist.

Lateral reading: leave the first page

Most of us read vertically. We stay on the article, scroll, and let the page persuade us with its own layout, quotes, and about box.

Lateral reading — documented by the Stanford History Education Group’s Civic Online Reasoning work (Wineburg, McGrew, and colleagues) — is what professional fact-checkers tend to do instead:

1. Leave the original page. Open new tabs.
2. Investigate the source. Search the outlet and author elsewhere.
3. Find independent coverage. Look for reporting that does not copy the same wording.
4. Trace the claim to its origin. Who said it first? Is it a press release, a social post, a rumour, or a primary document?

The point is not cynicism. It is provenance. A polished page can still be a closed loop.

A useful habit: if you have spent more than a minute only on the original URL, you are still inside the source’s story about itself.

Images, video, and AI-generated media

Pictures travel without their captions. In the course, Maya’s viral photo pre-dates the claim. That is a common pattern: a real image, a new story.

Practical checks:

• Reverse-image search (Google Images, TinEye, or similar). Ask: has this photo appeared before, in another place or year?
• Cropping. What was cut out — a sign, a date, other people, a wider scene that changes the meaning?
• Miscaptioning. Does the filename, EXIF data if available, or earlier publication match the new caption?
• Video. Look for mismatched audio, reused B-roll, and captions that assert more than the footage shows.

What AI detectors cannot establish

Synthetic media and deepfakes are a real risk. Tools that claim to “detect AI” can be useful as one weak signal. They cannot prove authenticity, and they can misfire on heavily compressed, edited, or unusual genuine footage.

Do not outsource the decision to a detector score. Combine provenance (where did this file first appear?), context (does independent reporting exist?), and humility (if you cannot establish origin, say so).

A five-step workflow you can reuse

This is the spine of the course, and it is meant for the next link you would actually forward.

1. Pause. Do not let the share button finish your thinking.
2. Name the precise claim. One checkable sentence. Not a vibe. Not a thread.
3. Trace the evidence. Who said it, when, with what document, photo, or data?
4. Compare credible sources. Look for independent coverage, not six copies of the same post.
5. Assign a confidence level before you post.

Confidence language is part of the job:

• High: multiple independent sources, clear origin, media checks out.
• Medium: some support, important gaps remain.
• Low: origin unclear, single source, emotional framing, or reused media.

A bounded update is often the professional move: Here is what is confirmed, here is what is still unverified, here is what we are not repeating. That is more useful than silence, and safer than a rumour with your name on it.

Activity 3
The next time a headline or photo creates urgency in a team or family chat, run the five steps on paper or in a notes app before you reply. Share the confidence level, not only the link.

Maya’s lesson — and yours

Maya’s story ends when she ships a bounded update instead of an unofficial incident report. The walk-back path is also realistic: a fast forward becomes the version everyone remembers, even after it is quietly corrected.

The transferable lesson is small enough to keep:

Name the claim type. Slow the spread. Audit before you cite. Leave the first page. Verify the media, not the mood. Then assign confidence.

Your next forward is the practice.

Further reading

• Wardle, C. and colleagues, First Draft / information disorder framework — definitions of misinformation, disinformation, and malinformation.
• Stanford History Education Group, Civic Online Reasoning — lateral reading and source investigation.
• News Literacy Project and similar public checklists — author, evidence, date, corroboration, and persuasion cues.
• Vosoughi, S., Roy, D. and Aral, S. (2018) ‘The spread of true and false news online’, Science.
• Public guidance on reverse-image search, and on the limits of AI-generated-media detectors.

Written by Ian Tappan, Co-Founder of CourseEngine.ai