How to control what ChatGPT says about you
You cannot edit the model and there is no form. What you can change is what it finds when it goes looking, and that turns out to be most of what matters.
Somebody asks ChatGPT who you are. It answers. You had no part in it.
This is the question I get most often now, usually phrased as a complaint, and it deserves a straight answer rather than a pitch. So here is the straight answer first: you cannot edit the model, there is no form, and anybody who tells you otherwise is selling something. What you can do is change what it finds when it goes looking, and that turns out to be most of what matters.
Where an answer about you actually comes from
Two places, and they work differently.
The first is training. The model absorbed a large amount of text at some point in the past and some of it was about you, or about someone with your name. That part is frozen, it is old, and you have no access to it. If it is wrong, waiting is the only lever, and the wait is measured in model releases.
The second is retrieval, and this is the one people miss. When an assistant answers a question about a named person today, it usually searches first and then writes from what it found. ChatGPT search and Copilot lean on Bing for that. Others use their own crawler or a third-party index. The words in the answer are frequently lifted from a page that existed this morning.
That second path is live, and it is the whole lever. You are not editing a model. You are giving the retrieval step something better to find than what it is finding now.
What actually gets quoted
Having a page is not the same as having a page an answer engine will use. The ones that get quoted tend to share four properties, and the fourth is the one most people fail.
It is readable without running anything. If your words only exist after a browser executes JavaScript, a good share of crawlers see an empty document. This is the quiet failure mode of most personal sites built in the last five years.
It is attributed to a named person. An answer engine is more willing to quote a claim it can attribute than a floating sentence. Your name on the page, in the markup as well as in the design.
It is specific. "Experienced leader with a passion for building teams" is not quotable, because it says nothing that could be false. "I stopped running design reviews on Fridays because nobody argues on a Friday" is quotable. Specificity is not a style preference here, it is the difference between being cited and being skipped.
It is reachable. This is the one people fail. LinkedIn sits behind a login wall for most crawlers, so the profile you have spent years on is close to invisible to the systems now answering questions about you. The same is true of anything that lives only in a PDF, a slide deck, or someone else's platform.
Go and check, right now
Before you do anything, ask ChatGPT and one other assistant who you are. Use your full name. Then ask something narrower, the thing you would want a client to hear, like what you think about hiring or how you run a migration.
Read what comes back and notice which of three things happened. It confused you with someone else, which is a retrieval problem and is fixable. It gave a vague summary with no source, which means it found nothing better and is very fixable. Or it quoted something real and attributed it, which means the path already works and you want more on it.
Most people who run this test for the first time get the second outcome. That is good news. A vague answer means nothing specific was available, not that something specific was rejected.
What to do about it
Write down the things you actually think, in your own words, somewhere a crawler can read without a login. Ten things is enough to change what a search returns. They should be the opinions you would defend in a meeting, not a biography.
Then make it findable. Link to it from the profiles you already have, because corroboration across several places is a large part of how these systems decide a page is really about the person it names. Your site, your signature, your LinkedIn featured section, wherever your name already appears.
Then wait, and check again. This takes weeks, not hours. Indexes update on their own schedule and no amount of wanting it changes that.
Where we come into it
This is what Humble and Kind is, so I will say it once and let you judge whether it is useful. You write down what you know. Your page answers questions from it, quotes your own words back with sources, and says plainly that it is an AI and not you. It is readable without JavaScript, it names you in its markup, and it publishes a plain-text version specifically so an assistant can read it in one fetch.
You could build all of that yourself. Several people have, and their pages work fine. The part that is harder to build alone is the part where you decide who gets which answer, so that a stranger, a client and a recruiter are not all handed the same thing.
But the advice above stands whether or not you ever use this. Write specific things. Put them where a crawler can reach them. Check what comes back. The people who do that will be quoted, and the people who keep everything on a profile behind a login will keep wondering why the machine says something vague about them.