You cannot edit what an AI assistant says about your brand. There is no correction form for businesses and no login that rewrites an answer for everyone. The fix runs through the sources the assistant reads: find the wrong fact, find the page or listing it came from, correct that source, make sure the crawlers can read the correction, report the answer through the assistant's feedback control, and retest the same prompt until the answer changes. OpenAI's help center says that when ChatGPT gives an incorrect answer, you should report it with the thumbs-down button under the answer. That report is feedback, not a guaranteed correction, so the source fix is the part you control.
Why the wrong answer exists
- A stale page. Your site still shows a retired price, a discontinued product, or an old address, and the assistant repeats it.
- Conflicting pages. Your homepage, a product page, and a directory listing give three different answers, and the assistant picks one.
- A similar name. Another brand shares your name or your old name, and the assistant merges the two.
- Too little published. When a fact appears nowhere clear on the web, the model fills the gap, and the fill is often wrong.
- A blocked crawler. Your correction is live but the search crawler cannot fetch it, so the assistant keeps reading the old copy.
- Training data. An answer built from training rather than live search changes only when a newer model arrives, so a source fix will not move it right away.
The correction sequence
1. Capture the wrong answer exactly
Save the prompt, the full answer, the date you saw it, and every citation or link shown. Note which assistant gave it. Test the same prompt in other assistants too, since the same wrong fact often appears in more than one place, and each one needs its own report.
2. Work out whether it came from search or training
If the answer shows citations or links to live pages, it came from search, and fixing those pages can change it after the next crawl. If it shows no citations and reads like remembered knowledge, it likely came from training, and you should still fix the sources, but expect the change to wait for a newer model. OpenAI's own help pages warn that search results and citations can be incomplete, outdated, or incorrect, so a cited source is a lead, not proof the source is right.
3. Fix your own pages first
Correct the fact on every page you control: the homepage, the product or service page, the About page, pricing, and the FAQ. Make the corrected fact easy to find in plain text, and make the structured data match it. A price in your markup that still shows the old number will keep the wrong answer alive.
4. Fix the listings and directories next
Check your Google Business Profile, Bing Places, and the review sites and directories the assistant cites. Make the name, address, hours, and product facts identical everywhere. Where a listing belongs to someone else, ask the site owner to correct it and give them the link to your corrected page.
5. Make sure the search crawler can read the correction
Allow OAI-SearchBot in robots.txt if you want the corrected page in ChatGPT search answers. OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, and that robots.txt changes take about 24 hours to take effect for search. GPTBot is the separate training crawler, so allowing one does not automatically allow the other. Check the same allow rules for the other assistants you tested.
6. Report the answer, after the source is fixed
Use the feedback control under the wrong answer in each assistant. For ChatGPT, that is the thumbs-down button. Include three things: the exact wrong sentence, the correct fact, and the link to the page that proves it. Report it once per assistant. A report about a fact that is still wrong on the web changes little, which is why this step comes after the fix.
7. Retest the same prompts on a schedule
Ask the same questions again with the same wording, and log the answer and the citations each time. Search-based answers can move after a recrawl. Training-based answers will not move until a model update, so keep the log and stop expecting a change between model releases. If the wrong fact is personal data about a named person rather than a company fact, check OpenAI's privacy request pages for a separate route, since that route covers personal data, not general company facts.
Say you run an outdoor gear brand that retired a tent line
A customer asks ChatGPT for your lightest tent, and it names a model you stopped selling, citing your old product page and a retailer's listing. You remove the retired model from your catalog page, publish a short note on the page that replaced it, and ask the retailer to update its listing. You check that OAI-SearchBot can fetch the updated page. Then you report the wrong answer with the thumbs-down button, quoting the retired model name, the correct current model, and the link to your catalog. You retest the same prompt and log each result. The search-based answer changes after the next crawl. Any training-based version waits for a newer model.
What this guide does not cover
This guide covers correcting a wrong fact. It does not cover removing a listing, a legal dispute over a false claim, or the monitoring work of catching the next wrong answer early. If an answer is defamatory, bring in your legal team rather than treating a feedback button as the remedy.
Check your own brand now
Run the free scan. It takes seconds, no signup, and shows whether AI crawlers can read your corrected pages today: aizii.ai/aeo-scan
Run the free scanSources
- OpenAI bots documentation (specific): https://platform.openai.com/docs/bots
- OpenAI Help Center, What is ChatGPT (specific): https://help.openai.com/en/articles/6783457-what-is-chatgpt
- Google Search Central, AI features and your website (specific): https://developers.google.com/search/docs/appearance/ai-features
- Google Business Profile Help, Guidelines for representing your business on Google (general docs): https://support.google.com/business/answer/3038177
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.
