Mental Avalanche

Using AI to detect FAKE NEWS

A Better Fake News Test: Don’t Start with Bias — Start with Deception

“Fake news” has become one of the most overused and least useful phrases in public debate. Too often, people use it to mean: “news I disagree with,” “news from the other side,” “news outside the mainstream,” or “news that has not yet been fully corroborated.”

That is a problem.

If everything controversial becomes “fake news,” the term loses its meaning. Worse, it becomes a tool for shutting down debate rather than improving public understanding.

A better approach is to define fake news more narrowly and more rigorously.

The key question should not be: Do I agree with this channel? Nor even: Is every claim fully proven?

The better first question is:

Is the channel deceiving viewers about what they are seeing, hearing, or being promised?

That is the heart of a deception-based fake news test.

Fake news is not the same as bias

A channel can be opinionated without being fake news.

It can be left-wing, right-wing, anti-war, pro-Western, anti-Western, religious, spiritual, anti-establishment, libertarian, socialist, conservative, mainstream or non-mainstream. None of that automatically makes it fake news.

Strong viewpoints are not the problem. Deception is the problem.

A political channel may be openly ideological and still be honest about what it is presenting. A health commentator may offer controversial interpretations and still be speaking under their own name, using real documents, and not fabricating evidence. A spiritual teacher may make claims many people reject, but that does not automatically mean they are using fake media or impersonation.

The deception test asks something narrower:

Is the viewer being tricked?

The strict deception test

Under this approach, fake news means there is clear evidence of deceptive presentation practices.

That includes five main categories.

1. AI impersonation or synthetic deception

This is one of the clearest modern forms of fake news.

Red flags include:

This is a hard fail.

If a channel creates an AI version of a doctor, journalist, politician, academic or celebrity and presents it as real, that is not simply “alternative media.” It is deception.

2. Fabricated or manipulated media

Another clear category is fake or manipulated evidence.

This includes:

This matters because video and screenshots carry authority. People often believe what they see before they check where it came from. A fake screenshot or altered clip can travel faster than a correction.

3. False title, thumbnail or content mismatch

This is one of the most common forms of deception on YouTube and social media.

The question is simple:

Does the video deliver what the title or thumbnail promises?

If the title says someone “admits,” “confesses,” “exposes,” “destroys,” “breaks silence,” or “reveals the truth,” did that actually happen?

Or is the video just speculation, commentary, or a weak reinterpretation of something else?

Sensational language alone is not enough to call something fake news. But repeated title-content mismatch is a serious red flag. If the title promises an explosive revelation and the video never delivers it, the viewer has been misled.

4. False context

Sometimes the footage is real, but the context is fake.

Examples include:

This is especially dangerous because it uses real material to create a false impression. The deception is not in the pixels; it is in the framing.

5. Hidden or fake identity linked to authority

Anonymous channels are not automatically fake news. But identity becomes relevant when a channel falsely claims authority.

Red flags include:

The issue is not anonymity by itself. The issue is false authority.

What this test deliberately excludes

This test does not ask whether a channel is politically neutral.

It does not ask whether mainstream fact-checkers agree with the channel.

It does not ask whether the channel’s interpretation is correct.

It does not ask whether the content is controversial.

It does not ask whether the presenter is mainstream, anti-mainstream, activist, ideological or one-sided.

Those are separate questions.

They may matter for media literacy, but they are not the same as fake news.

A channel can be biased but not fake. A channel can be controversial but not fake. A channel can be wrong without being deceptive. A channel can require verification without being fake news.

This distinction matters because it keeps the test fair.

The verdict scale

A practical fake news detector should give a clear result.

✅ PASS — Not fake news

Use this when there is no clear evidence of:

A pass does not mean “everything this channel says is true.” It means there is no clear evidence that the channel uses fake-news deception tactics.

⚠️ CAUTION — Check before sharing

Use this when there is some evidence of misleading presentation, but not enough to call the channel fake news.

Examples:

This is the middle category. It says: do not dismiss automatically, but check carefully before sharing.

❌ FAIL — Fake-news-like or deceptive

Use this when there is clear evidence of:

This is the category for channels that actively deceive viewers about what they are seeing or hearing.

The triple quality-control check

Before giving a verdict, apply three checks.

QC 1 — Scope check

Am I judging deception only?

Or have I drifted into judging ideology, political position, scientific interpretation, controversy, or whether I personally agree?

If the answer is yes, reset the analysis.

QC 2 — Evidence check

Have I found clear evidence of deceptive presentation?

Or am I assuming deception because the channel is controversial, emotional, one-sided, or outside the mainstream?

If there is no clear evidence of deception, do not downgrade.

QC 3 — Verdict check

If there is no clear evidence of the defined deceptive practices, the verdict must be:

✅ PASS — Not fake news

This is the discipline of the test.

Why this approach is useful

This approach does not solve every media literacy problem. It does not tell us whether a channel is balanced, accurate, rigorous or wise. It does not replace deeper fact-checking.

But it does something important.

It separates deception from disagreement.

That makes the fake news label harder to abuse.

It also helps us deal with the modern information environment, where the biggest risks are not only bad arguments, but fake identities, synthetic voices, misleading thumbnails, false context and fabricated authority.

In other words, the first test should not be:

Do I like this source?

The first test should be:

Is this source tricking me?

If the answer is no, it may still need verification. It may still be biased. It may still be wrong. But it should not automatically be called fake news.

That distinction is essential if we want a healthier public debate.