How do I detect when a public AI model states a wrong fact about our company?
By SensoPublished Last updated 4 min read
List the facts about your company that must be right, ask public AI models the questions that surface those facts, and compare every claim in each answer against your own source documents. Flag any claim that contradicts your documents, and send any claim your documents do not mention to the person who owns that fact.
A wrong fact in an AI answer is a claim about your company that conflicts with your own approved documents, such as a price, feature, policy or date. Senso by Zeta, the generative engine optimization (GEO) platform from Zeta Global, runs this comparison automatically and marks each claim confirmed, contradicted or not covered.
How often do AI models state wrong facts?
Often enough to check regularly. In an October 2025 study led by the European Broadcasting Union and the BBC, journalists from 22 public service media organizations in 18 countries evaluated more than 3,000 answers from ChatGPT, Copilot, Gemini and Perplexity. About a fifth contained major accuracy issues, such as hallucinated or outdated information. The study tested news questions, not company questions, so your own rate can only come from checking your own answers.
Which facts should you check first?
Start with the facts buyers use to compare you with other options.
| Fact type | Examples | Why check it first |
|---|---|---|
| Product and pricing | Product names, plans, prices | Buyers use these to judge cost |
| Capabilities | Key features and integrations | Buyers use these to build a shortlist |
| Trust | Security and compliance claims | Buyers use these in security and procurement reviews |
| Company | Leadership and company history | Buyers use these to judge credibility |
How do you detect a wrong fact step by step?
Four steps take you from a fact list to a set of flagged claims.
| Step | Input | Output |
|---|---|---|
| 1. Write questions | Your fact list | Questions that would surface each fact |
| 2. Run them | The public AI models you track | Recorded answers |
| 3. Compare claims | Each answer and your source documents | A matching, conflicting or missing passage per claim |
| 4. Flag | The comparison | Claims that contradict your documents, and claims your documents do not mention |
A claim your documents do not mention is not necessarily wrong. The person who owns that fact needs to confirm whether it is true.
When should you repeat the check?
Repeat on a fixed schedule and after every change to a fact, because AI can keep repeating the old version for some time.
| Trigger | Why |
|---|---|
| A fixed schedule | Answers change between runs |
| A price change | AI can keep repeating the old price |
| A product launch | AI can keep describing the previous product |
What mistakes let wrong facts slip through?
Most missed errors look almost right.
| Mistake | Why it fails | Do this instead |
|---|---|---|
| Skimming the answer as a whole | A single wrong detail hides in a mostly right sentence | Compare each claim on its own |
| Checking only after complaints | Buyers may never tell you what AI said | Check on a schedule |
| Treating an unmentioned claim as false | Your documents may be missing a true fact | Ask the owner of that fact |
| Checking one model | Each model can get a different fact wrong | Check every model your buyers use |
How does Senso by Zeta detect a wrong fact about your company?
Senso by Zeta asks AI models the questions people in your market ask, on a schedule, and fact-checks the answers that name you. The fact-check splits each answer into claims and checks each one against your Context Layer, the facts your organization stands behind.
| Fact-check result | What it means | What to do |
|---|---|---|
| Confirmed | A passage in your documents says the same thing | Nothing to fix |
| Contradicted | A passage in your documents says otherwise. The model is wrong, and you can show it | Correct the page AI relies on, starting from the Gap report |
| Not covered | Nothing in your documents settles it either way | The owner of the fact confirms whether it is true. If it is, it goes into your Context Layer |
A wrong fact is often a sentence that sounds right with one detail wrong. The Senso documentation's example is an answer that gives a 60-day refund window when the refund policy says 30 days. Because tracked questions run on a schedule, a wrong fact that appears after a price change or launch shows up in the next run.
What results have companies seen with Senso by Zeta?
FundMore.ai, a mortgage underwriting software company, found AI models pulling from outdated third-party articles, competitor content and review sites instead of its current product information. Using Senso, it launched a dedicated domain, fundmore.finance, on April 8, 2026. By April 20, 2026, that domain was one of the top 10 most-cited external sources across its tracked questions, with 75 citations and a 60% average share of voice (FundMore.ai case study).
Sources: EBU and BBC, News Integrity in AI Assistants, October 2025. Senso documentation, Why Senso. Senso documentation, Verification Loop.
Next step: Check what AI models say about your company with the Senso by Zeta Verification Loop.
Published by Senso by Zeta, the GEO platform from Zeta Global.
