Aizii Learn · AI Search

[← Back to Learn](https://aizii.ai/learn)

# How to set up alerts for AI misinformation about your brand

An alert for AI misinformation is a tripwire: a check that tells you when an assistant starts saying something wrong about your brand, so you can fix the source before many shoppers hear it. There is no single alert that watches every assistant, and no vendor offers a built in alarm for wrong answers about your brand. The setup in this guide has three parts that work together: web alerts that catch new pages carrying wrong facts, a regular prompt check that catches wrong answers directly, and a triage routine that turns an alert into a source fix. An alert that nobody reads is not a tripwire. The routine matters more than the tool.

Start by deciding what counts as misinformation worth an alert. A wrong price, a discontinued product presented as current, a false claim about ingredients or materials, and a wrong address or contact detail all count, because each one can send a shopper to the wrong decision. A mildly unflattering opinion does not count, and a competitor being named instead of you is a visibility problem, not misinformation. Write the list down before you set up any alert, because the list decides the search terms, and the search terms decide what you hear about.

## Where it goes wrong

Alert setups fail because they watch the wrong thing, or because nobody acts on what they catch.

-   The alert watches the brand name only. A wrong price on a product page never mentions the brand name in the alert terms, so the alert stays silent while the wrong answer spreads.
-   The alert watches the whole web and drowns the owner. A broad term with no quotes and no exclusions returns hundreds of results a week, and the owner stops reading them. An unread alert is the same as no alert.
-   The alert catches web pages, not AI answers. Google Alerts tells you when a new page appears in search results. It does not tell you what an assistant said in an answer. A setup with web alerts only will miss a wrong answer that cites an old page.
-   The prompt check has no fixed wording. Each check uses a different phrasing, so a changed answer cannot be tied to a changed source. The check needs the same wording every time, or it measures noise.
-   The alert fires, and the fix goes to the wrong place. The owner reports the answer to the assistant before fixing the source page, and the wrong fact comes back the next time the assistant reads the same page. The source fix comes first, and the report comes after.
-   The alert has no owner and no response time. An alert that arrives in a shared inbox with no named owner waits until someone happens to read it. A wrong price can answer shoppers for weeks while the alert sits unread.
-   The tool is trusted for what it does not do. A social listening tool watches social networks, news, blogs, and forums. It does not watch what an assistant says in a private answer. A tool that is described as monitoring mentions is not automatically monitoring AI answers, so check what each tool actually watches before you rely on it.

## The steps

**Step 1. Write the alert terms from your misinformation list.** Build one term for your brand name, one for each product name that carries a price or a claim shoppers act on, and one for each wrong fact you have already seen, such as a discontinued model or an old price. Use quotes around exact phrases, so the alert matches the phrase and not each word on its own. Add an exclusion for the competitor with a similar name, so their news does not fill your inbox. Keep the list short: five to ten terms you will actually read beats thirty terms you will ignore.

**Step 2. Set up web alerts for new pages.** Google Alerts sends an email when new results for a topic show up in Google Search. Create one alert per term from step 1, and choose the settings Google offers for how often to receive them and which results to include. Check the vendor docs in the Sources section for the current settings before you rely on a menu name in this guide. Google Alerts watches search results, not AI answers, so treat it as the tripwire for new source pages, not as a watcher of assistants. A new retailer listing with the wrong price is the kind of thing it catches well.

**Step 3. Add a prompt check that watches the answers directly.** Pick the assistants your shoppers use, and pick the prompts from your monitoring log that carry the facts shoppers act on: price, availability, ingredients or materials, and the product that replaced a discontinued model. Ask each prompt with the same wording, in a fresh conversation, and save the full answer and every citation. Run the check once a week, and more often around a launch or a price change. This is the part of the setup that catches a wrong answer, and no web alert replaces it.

**Step 4. Add a paid visibility tool only if you need its coverage, and check what it watches.** Profound is a tool that tracks how brands appear in AI answers, with visibility, citation, and sentiment reports across answer engines, according to its own documentation. Check its own site for current plans and for what its alert options cover before you buy it, since plans and features change. Do not assume a visibility tool watches every assistant, or that it alerts on every wrong fact. Use it for the coverage your prompt check cannot reach, such as many prompts across many engines, and keep the prompt check for the facts that matter most.

**Step 5. Give every alert an owner and a response time.** Name one person who reads the alerts, and agree how fast a wrong price or a discontinued product gets a source fix. A wrong price that sends shoppers to a dead end is urgent. A wrong description on a low traffic page can wait for the next working day. Write the two levels down, so the owner does not have to decide under pressure. Put the alerts in one place, such as a shared inbox or a channel the owner checks daily, and never in a personal inbox nobody else can see.

**Step 6. Triage every alert the same way.** When an alert fires, save the prompt or the page, open the citation, and find the source page that carries the wrong fact. Fix your own pages first, then ask other site owners to fix theirs. Report the wrong answer through the assistant's feedback control only after the source is fixed, with the exact wrong sentence, the correct fact, and the link to the corrected page. Log the alert, the source, the fix, and the retest result in the same log you use for monitoring. An alert with no log entry will be investigated twice.

**Step 7. Review the alert terms every month.** Read the alerts that fired and the ones that did not. Drop a term that only returns noise, add a term for each new wrong fact you found, and add a term for each new product you launch. Retire the terms for products you discontinued, or the alerts will keep firing for a product you no longer sell. A monthly review keeps the setup quiet enough to read and sharp enough to catch the next wrong fact.

## The machine-readable piece

This alert term list is the setup from step 1, written as plain text you can copy into Google Alerts, one term per alert. The brand and product names are placeholders: replace them with your real names, and keep the quotes around exact phrases. The exclusion line keeps a similarly named competitor out of your results. Check the Google Alerts help page in the Sources section for the current settings before you create each alert.

```
"[yourbrand]"
"[yourbrand] [your main product]"
"[discontinued product name]"
"[yourbrand] price"
"[yourbrand]" -[competitor name]
```

This triage record is your own log format, not a vendor format. No vendor defines how you should log an AI misinformation alert, so this file is a working template you can copy into a spreadsheet. One row per alert, with the source page that lets you fix it.

```
alert_term,assistant_or_source,wrong_fact,correct_fact,source_url,fix_done,retest_result
"[yourbrand] price",ChatGPT,"old price shown","current price","https://[yourdomain.com]/products/example-product","price fixed on page and in markup","correct on retest"
"[discontinued product name]",Google Alerts,"retailer lists retired model","model discontinued","https://[retailer.example]/example-product","retailer asked to update","waiting on retailer"
```

## What it looks like when it works

On the page, nothing looks different to a human reader. The change is in the inbox and the log: a web alert arrives when a new page carries a wrong fact, the weekly prompt check catches a wrong answer the web alert missed, and the triage routine turns each alert into a source fix within the agreed response time. A wrong price is fixed at the source, reported to the assistant, and confirmed correct on retest, all logged in one row. The monthly review shows fewer repeat alerts for the same fact, because the source fix held, and the alert terms stay short enough that the owner reads every one.

Say you run a supplement brand, and an assistant starts telling shoppers your protein powder contains an ingredient you removed. Your web alert for the product name fires when a retailer publishes an old product page with the removed ingredient. Your weekly prompt check catches the same wrong answer in ChatGPT, citing that retailer page. You fix your own product page first, then ask the retailer to update its listing, sending the link to your corrected page. You report the wrong answer through the assistant's feedback control after the source is fixed, with the exact wrong sentence and the corrected page. You log the alert, the fix, and the retest. The next round of checks shows the correct ingredient list, and the monthly review adds the old ingredient name as an alert term, so the next stale listing fires an alert the day it appears.

### The fundamentals

These mechanics are shared across every engine and covered in full in their own guides:

-   •The llms.txt format and a complete sample, see our [guide to writing a good llms.txt file](https://aizii.ai/learn/ai-search/how-to-write-a-good-llms-txt-file)
-   •The markup AI systems read to verify facts, see our [product schema guide](https://aizii.ai/learn/ai-search/how-to-add-product-schema-markup-for-ai-search)
-   •The complete crawler rule reference, see our [robots.txt guide for AI search](https://aizii.ai/learn/ai-search/how-to-write-robots-txt-for-ai-search)

## What this guide does not cover

This guide covers setting up alerts for AI misinformation about your brand, with web alerts, a prompt check, and a triage routine. It does not cover correcting a wrong answer in full, which is a different job with its own guide, or removing outdated product info, which has its own guide too. It does not cover legal removal of defamatory or personal information, which uses separate request routes. It does not cover social listening for what people say about your brand on social networks, which watches a different kind of mention. Those are separate jobs, and this guide is the tripwire, not the whole program.

## Check your own store now

Run the free scan. One click, no signup, and it shows whether AI crawlers can read your store today: aizii.ai/aeo-scan

[Run the free scan](https://aizii.ai/aeo-audit)

## Sources

-   Google Search Help, Create an alert (specific): [https://support.google.com/websearch/answer/4815696](https://support.google.com/websearch/answer/4815696)
-   OpenAI, Overview of OpenAI Crawlers (specific): [https://platform.openai.com/docs/bots](https://platform.openai.com/docs/bots)
-   Google Search Central, AI features and your website (specific): [https://developers.google.com/search/docs/appearance/ai-features](https://developers.google.com/search/docs/appearance/ai-features)
-   Profound, Analytics and reporting (general docs): [https://docs.tryprofound.com/mcp/capabilities/analytics-capabilities](https://docs.tryprofound.com/mcp/capabilities/analytics-capabilities)
-   Google Search Central, Product structured data (specific): [https://developers.google.com/search/docs/appearance/structured-data/product](https://developers.google.com/search/docs/appearance/structured-data/product)

Note: links marked specific point to the page that supports the claim in this guide. Links marked general docs point to the vendor's general docs section for that topic, not to a page that proves every step. Check the vendor docs for the current path before relying on a menu name or click path in this guide.
