Judging Source Credibility

Detecting AI text and images and catching hallucinations all lead to the same deeper question: should you trust this source at all? Whether a claim comes from a website, a social account, a news outlet, or an AI tool's cited reference, you need a fast, repeatable way to size up whether the source has earned your trust. This lesson gives you one.

Credibility is not a yes-or-no badge. It is a judgment you make from a few signals, and it gets stronger when independent sources agree. The goal is a quick read you can run in under a minute, plus a deeper check for the claims that really matter.

What You'll Learn

• The three pillars of source credibility: authority, corroboration, and recency
• A faster technique that professional fact-checkers actually use
• Common red flags that should drop your trust quickly
• A short checklist you can apply anywhere

The Three Pillars

When you need to judge a source, weigh three things together. No single one is decisive; they reinforce each other.

Authority asks whether the source is in a position to know. A named subject-matter expert, an official agency, a primary document, or an established outlet with editorial standards outranks an anonymous post or a site with no accountability. Watch for borrowed authority too: a real expert quoted out of context, or credentials that sound impressive but are unrelated to the topic.

Corroboration is the strongest of the three, because it is the hardest to fake. One source making a dramatic claim is just a claim. Several independent sources reporting the same thing is much closer to a fact. The word independent matters: ten sites all copying one original post are not corroboration, they are an echo. Trace claims back and check whether the agreement is real or just repetition.

Recency asks whether the information is current. On fast-moving topics (prices, technology, laws, ongoing events), an old page can be confidently wrong simply because the world changed. Check publication and update dates, and be wary of undated content presented as current.

Lateral Reading: What Fact-Checkers Actually Do

Most people evaluate a source by staying on it: reading its About page, judging its design, seeing if it looks professional. Research into professional fact-checkers found they do the opposite, and do it faster. The technique is called lateral reading: instead of going deeper into the page in front of you, you open new tabs and find out what other, independent sources say about it.

To read laterally:

• Leave the page. Before trusting a site or account, search its name elsewhere. What do independent outlets, reference sources, and critics say about it?
• Check who is behind it. Look up the organization or author. Who funds them? What is their track record and agenda? A polished site can still be a front for a single interest.
• Confirm the specific claim elsewhere. Search the claim itself, ideally back to its primary source, rather than accepting the version on the page.

Lateral reading works because a site's own appearance is fully under its control, and therefore tells you little. What independent sources say about it is much harder to manufacture. This is the single most valuable credibility skill in this lesson.

Common Red Flags

Some signals should drop your trust quickly, even before a full check:

• No accountability. No named author, no real organization, no way to contact anyone, no corrections policy.
• Emotional manipulation. Content engineered to enrage or terrify you, often with urgent calls to share immediately. Strong emotion plus pressure to act fast is a classic manipulation pattern.
• A clear undisclosed motive. A "review" that exists to sell you something, or "news" from a source that profits from the conclusion.
• Mismatched or missing sourcing. Big claims with no citations, or citations that, when you check, do not say what the page claims they say.
• Imitation. Accounts and sites mimicking a trusted brand or outlet with a slightly-off name, logo, or web address.
• Can't be corroborated. A major claim that should be widely reported but appears only in one obscure place.

None of these alone proves a source is wrong, but together they justify strong skepticism, and any one of them is a reason to corroborate before believing.

A Quick Credibility Checklist

Run this fast read on any source that matters:

1. Who, and how would they know? Named, accountable, and actually positioned to know beats anonymous.
2. Read laterally. Open a tab and see what independent sources say about the source and the claim.
3. Corroborate. Find at least one or two genuinely independent sources that agree. Watch for echoes versus real agreement.
4. Check the date. Make sure it is current enough for the topic.
5. Scan for red flags. Emotion, undisclosed motive, no accountability, imitation, missing sourcing.

Apply this to AI output too. When a chatbot gives you a source, the same checklist applies to that source, and the corroboration step is exactly how you catch a hallucinated citation that does not hold up.

Key Takeaways

• Judge sources on three pillars working together: authority (who, and how would they know), corroboration (independent agreement), and recency (current enough to be right).
• Corroboration is the strongest pillar, but only when the sources are genuinely independent, not echoes of one origin.
• Read laterally: leave the page and see what independent sources say about it, rather than judging it by its own appearance.
• Treat emotional manipulation, undisclosed motive, no accountability, and imitation as red flags that demand corroboration before trust.