VARNIUMAI risk defense / everyday bearings

DEEPFAKE / QUOTE VERIFICATION

Verify a Quote Before You Reshare: The AI-Generated News Check.

A quote can be fabricated, misattributed, or generated wholesale and attached to a real name. The check verifies the source before the quote becomes your post.

Start here

The free tool below runs entirely in this browser — no account, nothing stored or transmitted. Redact anything you would not want kept before you type it.

Open the companion tool →

Why wording is the weakest evidence

Generators are fluent with words, which means a fabricated quote reads exactly like a real one. The only check that survives is the source: does this quote exist in the person's own words, on their verified account, in a transcript, or in a reputable publication.

A misattributed quote travels farther than a correction ever will. Verifying before you reshare is the entire defense, and it takes under a minute once it is a habit.

The protocol

  1. Enter the quote and who it is attributed to.
  2. Read the source reading: primary source, publication, and the independent confirm step.
  3. Search the exact quote on the person's verified account or a reputable outlet before sharing.
  4. If you cannot confirm it, hold the post and say so if someone asks.

What the tool does

Enter a quote and its attribution and get a source-verification reading with the basis shown. It runs in this browser; nothing is sent or stored.

Limit first: This check can raise your confidence, but a clean result doesn't prove a quote is real. It structures the source questions — it does not search the web for you.

Open the companion tool →

One human next step

Search the exact quote on the person's verified account or a reputable outlet. See CISA's disinformation tactics guide for the broader pattern.

Research log and safety checks

AI-assist path: not used for a verdict. This artifact is an inspectable source-verification aid; a remote model would add an opaque judgment where the reader needs to see the source gap. §6.10: not a frontier-pace topic.

§0.5 protection result: category is deepfake-media-verification; crisis-adjacent: no; fear/urgency/scarcity toolkit used: no. The protective function is free and the paid feature is convenience only.

Related dispatches

Pairs with the same protection bar: a free complete reading, an optional convenience unlock, and no fear-framing.