InstaFact ← Back to site

Editorial

Content & Corrections Policy

Effective 3 September 2026 · Last updated 3 September 2026

A verdict without a source is just another opinion. This page explains how InstaFact reaches a verdict, what it will not do, and exactly how to get one corrected when we have it wrong — which, being an automated system, we sometimes will.

1. Why this policy exists

InstaFact publishes assessments of claims made in public short-form videos. Assessments like that affect how people understand a topic, and sometimes how they see the person who made the claim. Publishing the method alongside the result is what makes the result checkable. This policy is that method.

2. How a check works

When you send a link, InstaFact:

  1. Retrieves the video from the public page you linked to.
  2. Transcribes the audio into text.
  3. Extracts the central claim — the specific, checkable assertion the video makes. Where a video makes several, we check the most prominent one.
  4. Searches for published evidence bearing on that claim.
  5. Weighs the evidence, giving more weight to primary research, systematic reviews and official statistics than to secondary reporting.
  6. Publishes a verdict with a confidence score, a plain-language explanation and the sources used.

These steps are automated. They are performed by software, including large language models, and a verdict is published without a human reviewing it first. Human review happens when a correction is requested, as described in section 9.

3. What we treat as a source

We prefer, in roughly this order:

We treat as weak or unusable:

Every source used is listed with the verdict and linked, so you can read what we read. Where the evidence is thin or contested, the verdict should say so rather than manufacture certainty.

4. The verdict scale

VerdictMeaning
TrueThe claim is supported by the available evidence.
FalseThe claim is contradicted by the available evidence.
MisleadingThe individual facts may be accurate, but the way they are presented leads to a false impression.
ExaggeratedThe claim points in the right direction but overstates the size, certainty or significance of the effect.
Hidden informationThe claim leaves out context that materially changes how it should be read.
UnverifiedWe could not find published evidence confirming the claim, though evidence may exist.
UnverifiableThe claim is not the kind of thing that can be checked against published evidence.

“Unverified” and “Unverifiable” are real outcomes, not failures. An absence of evidence is not evidence that a claim is false, and we will not mark a claim false merely because nobody has studied it.

5. What the score means

The confidence score reflects how consistently the sources we found supported the verdict — how much they agreed, how strong they were, and how directly they addressed the claim.

It is not a probability that the verdict is correct, and it is not a measure of how harmful a claim is. A high score on a claim where we retrieved the wrong evidence is still wrong.

6. What we do not do

7. Independence and funding

InstaFact is operated by LucidLabs. Verdicts are produced by the process in section 2 and are not influenced by any commercial relationship.

We do not accept payment from anyone to produce, alter, delay or withhold a verdict, and no advertiser, sponsor or third party has any say in what a check concludes. If that ever changes — if we take sponsorship or advertising — we will say so here before it takes effect.

8. Asking for a correction

Anyone can ask for a verdict to be reviewed — the person who made the claim, the person who requested the check, a cited publisher, or any reader who thinks we got it wrong.

Email connect@instafact.co with the subject line “Correction request” and include:

You do not need an account, and there is no charge.

9. How we handle it

  1. Acknowledgement within 48 hours.
  2. Human review. A person re-reads the claim, the verdict and the sources against the evidence you have sent.
  3. Outcome within 30 working days of receipt, one of:
    • Corrected — the verdict, explanation or sources are amended.
    • Clarified — the verdict stands, but wording is sharpened to remove ambiguity.
    • Withdrawn — the check is removed, where the claim should never have been assessed or the evidence cannot support any verdict.
    • Upheld — the verdict stands, with our reasoning explained to you.
  4. Transparency. A corrected or withdrawn verdict is marked as such, and the change is dated. We do not quietly rewrite a published verdict.

If you disagree with the outcome, reply to us with the reasons and we will look at it again. If we still reach the same conclusion, we will explain why and treat the matter as closed, without prejudice to any right you have to complain elsewhere.

10. Rights and takedown requests

If you hold rights in material we have referenced, or you believe a verdict contains unlawful content, write to connect@instafact.co with the specific URL, what the issue is, and your basis for raising it. Complaints under India's Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 are acknowledged within 24 hours and resolved within 15 days, and should be addressed to the Grievance Officer named in our Privacy Policy.

11. Contact

LucidLabs · connect@instafact.co


Related: Privacy Policy · Terms of Use