MEDIABIAS.fyi

Check a claim before you share it

Myth-busting

Twelve things people get wrong

These are myths about misinformation itself. Most of them are comforting, which is roughly why they persist.

Myth

Only gullible people fall for misinformation

Reality

Education and intelligence offer far less protection than people assume, and in some measured cases reasoning ability makes motivated reasoning more effective, because a better reasoner builds a better defence of a conclusion they already reached. Believing you are personally immune is one of the better predictors that you are not running the checks.

Myth

If a photo is real, the story is real

Reality

A genuine, unedited photograph from another country or another decade, relabelled, is among the most effective forms of misinformation in circulation. There is nothing wrong with the pixels, so looking harder at the image never helps. Only the caption is false, and reverse image search is the check.

Myth

You can spot AI-generated media if you look closely

Reality

In the Veriff Deepfakes Report 2026, a survey of 1,000 US adults fielded in February 2026, average accuracy at identifying manipulated video scored 0.07 on a scale from minus 1 to 1, where zero is guessing. Around half of respondents believed they could reliably tell the difference. Visual tells date quickly; provenance does not.

Myth

An AI detector settles whether something is AI

Reality

Detectors produce a probability with real error rates in both directions, and their performance falls on content unlike their training data. They flag human work as synthetic and pass synthetic work as human. Use one as a reason to investigate, never as the finding.

Myth

A .org domain means a nonprofit you can trust

Reality

Anyone can register a .org. Domain endings carry almost no verification, and neither does a professional design, a reference list, or a serious tone. All three are cheap. Funding is the thing that is hard to fake and easy to look up.

Myth

Lots of shares and comments mean it has been checked

Reality

Feeds rank by predicted engagement, and the most emotionally activating version of a story reliably outperforms the most complete one. Volume measures spread, not accuracy. Twenty accounts repeating one unsourced claim is one claim.

Myth

Reading a suspicious article carefully is the way to evaluate it

Reality

Professional fact-checkers do close to the opposite. They leave the page within seconds and look up the publisher elsewhere. Staying on the page means evaluating a source using only what that source chose to tell you about itself.

Myth

Screenshots are decent evidence that someone said something

Reality

A screenshot is an image of text, producible in a browser in under a minute with no special tools. It is one of the lowest-effort fabrications available and one of the most widely believed. Find the original post, or treat it as unverified.

Myth

Bias means an outlet is publishing false information

Reality

Most media bias involves no false statements at all. It is selection, framing, placement and omission: which fact leads, what things are called, whose reaction is quoted first, what is left out for space. Hunting only for false statements will find nothing and prove nothing.

Myth

There is a genuinely unbiased news source out there

Reality

Publishing requires choosing what to cover and what to lead with, and there is no neutral setting for those choices. The productive move is not finding a frameless source, it is reading two versions of the same event and noticing what each leaves out.

Myth

Correcting someone with facts changes their mind

Reality

It sometimes does, and it works far better when the correction engages the specific claim rather than the person, concedes any genuine grievance underneath, and is honest about uncertainty. Ridicule reliably hardens a position, because it converts a factual dispute into a status dispute. Explaining a technique before someone meets it works better than correcting after.

Myth

Fact-checking organisations are all the same, or all worthless

Reality

They differ substantially in funding, methodology and error-correction practice, and the label itself is not a credential since anyone can use it. Established organisations publish who pays for them, how they work, and what they got wrong. Applying the four moves to a fact-checker is consistency, not cynicism.