Misinformation

Sometimes an engine says something about your brand that is simply wrong: an old price, a product you discontinued, a founder who never worked there. The Truth screen compares the answers Rixe collects with the verified facts in your brand vault, raises a flag when an answer contradicts them, and takes each flag from "spotted" to "resolved" — with a correction plan, ready-to-paste feedback for the engines, and a watch period that checks whether the claim has really gone.

#Where to find it

Site › Presence & Surfaces, then the Truth tab (https://app.rixe.ai/presence?tab=truth).

#How flags are found

• The truth scan runs every night at 05:45 UTC over recent answers. Start contradiction scan runs it right away; flags show up within a few minutes.
• A model compares each claim about your brand with your verified facts. It is deliberately conservative: only contradictions it rates at 0.8 confidence or higher become flags. Telling you an engine is spreading a lie when it is not would cost your trust in every other number.
• Every automatic finding arrives as spotted and waits for a person. Nothing enters the correction process until you confirm it.

#Reading the screen

The top row shows Open flags (awaiting correction), Resolved, Median resolution (in days) and Most misinformation (the engine with the most flags). A line underneath names your oldest open claim — or says the engines describe your brand correctly.

Each flag card shows:

• the claim as the engine stated it, and its status;
• Engines: where it appeared, and Sources: the domains cited next to it (where it probably came from);
• Repeat rate: how often it is still repeated, and Last seen:;
• What's correct? — the true statement, which you write or edit. Correction texts are generated from this sentence.

#The life of a flag

A flag cannot jump from spotted to resolved; each step is a decision someone made.

#Step by step

1. Check the claim. Read it against your facts. If it is wrong, write the true version under What's correct? and press Confirm. If the engine was right, or it is not about you, press Not valid.

2. Create the correction plan. On a confirmed flag, press Create correction plan. Rixe creates:

   • one content brief for an FAQ-style page that states the truth plainly (it appears in Content);
   • one outreach task for each source domain carrying the claim, with instructions to ask for a correction (Sources, outreach and PR);
   • a JSON-LD FAQPage snippet so the fact is machine-readable.

   The message reads, for example, Correction plan ready: 1 content brief, 3 source tasks.

3. Report it to the engines (see below), then press Start correcting.

4. Watch. When the page is published and the sources are contacted, press Watch. Rixe keeps counting how often the claim comes back.

5. Resolve. A watched claim that has not been seen for 4 weeks is proposed as resolved. Press Resolved to close it. "Resolved" is only ever a suggestion: an engine's index can lag by weeks.

#Report to the engines

Fixing the sources is the lasting cure, but every engine also takes feedback on an individual answer. Once a flag is confirmed, being corrected or being watched, its card shows Report to the engines:

• a ready-made feedback text naming your brand, the claim and the correct information (plus your official source when your brand has a domain) — Copy text;
• for each engine where the claim appeared, Open engine and short instructions. For ChatGPT, Gemini, Perplexity, Copilot and Grok you ask the same question and use the thumbs-down under the answer; for Google you use the feedback link under the AI overview, and Suggest edits on your knowledge panel if you have claimed it.

If What's correct? is empty, the box asks you to fill it first — the feedback text is built from it.

#Alerts

When the same false claim appears on three or more engines, Rixe raises a False information is spreading alert, so you hear about it without opening this screen. See Alerts and inbox.

#Limits and plans

The truth scan, flags, correction plans and engine feedback are in every plan. On agency plans the nightly contradiction check draws credits. See Plans and limits.

#Common questions

Why is the list empty when I know an engine gets something wrong? Either the fact is not in your brand vault as a verified fact, or the model was not confident enough to flag it. Add the fact, then press Start contradiction scan.

Is "resolved" automatic? No. Rixe only proposes it after four clear weeks; a person closes the flag.

Does Rixe send the feedback to the engines for me? No. The engines take feedback from a person on an answer; Rixe prepares the text and the steps.

#Related

• Brand vault and knowledge base
• Mentions and perception
• Sources, outreach and PR
• Alerts and inbox