Can you control what ChatGPT says about you?
Not directly, and anyone promising you a guaranteed AI answer is selling something. But “no direct control” isn't “no control”. Assistants assemble their answer about you from sources: your profiles, your website, directories, press, structured data. Every one of those is something you can create, correct or improve. Control the inputs and the answer follows, usually within weeks to months rather than days.
There are two levers, and people constantly confuse them: removing what's wrong, and building what's right. You usually need both.
Lever 1: Correct or remove what's wrong
Start with the shortest paths to a fix:
- In your own account, wrong “facts” usually come from memory: open Settings, then Personalization, then Manage memories, and delete what's stale.
- For what ChatGPT tells other people, OpenAI runs a privacy portal at privacy.openai.com where you can request correction or removal of inaccurate personal information. Requests are reviewed case by case.
- Removal only mutes ChatGPT. If the wrong information lives on a website, other assistants keep reading it there, so chase the source too: ask the site owner for a correction, and update every profile you control.
Lever 2: Build the answer you want it to give
Assistants trust corroboration: the same facts, stated the same way, in several places they consider reliable. Concretely:
- Write one canonical bio (name, what you do, for whom, where) and use it everywhere: LinkedIn, X, Instagram, your site, speaker pages.
- Publish one page that states who you are as structured data (Schema.org Person markup), not just prose. This is the part machines actually parse.
- Link your accounts to each other so they corroborate: your site to your profiles, your profiles back to your site.
- Earn third-party mentions over time: directories, guest posts, podcasts, press. Corroboration you don't control counts the most.
- Keep it current. An assistant repeating your old job title is reading a page you forgot to update.
How long does it take to change your AI answer?
Weeks to months, not days. Assistants that browse (ChatGPT with search, Perplexity, Copilot) pick up improved sources within days to weeks of a re-crawl. Answers that come from training data move on the model's retraining schedule, which is why the browsing-based answer usually improves first and the offline answer catches up later.
The practical rhythm: fix the inputs once, properly, then re-run the check monthly instead of refreshing it daily. For the full check across assistants, use the 15-minute AI audit.
The shortcut
Search Presence does the structural half for you: it turns accounts you verifiably own into one machine-readable identity page, shows you what AI currently says about you, and keeps the two connected. The consistent bios and the third-party mentions stay your job; no honest tool can do those for you.