When someone asks ChatGPT, Perplexity or Google's AI Mode about you, the answer is put together from whatever the system has learned or can find about you online. If there is very little, the answer is vague or wrong. If the pages about you are out of date, so is the answer. Most of the fixes are the same as for ordinary search, with a few differences.
Where the answers come from
An assistant draws on two sources. The first is its training data, a large snapshot of the web taken at a fixed point in the past. Anything that changed after that date isn't in it. The second is a live search, run at the moment someone asks, which brings back current pages for the model to read and summarise.
Each company runs its own crawlers for the live part. OpenAI says its OAI-SearchBot crawler is used to surface websites in ChatGPT's search features, separately from GPTBot, which collects training data. Perplexity says PerplexityBot is designed to surface and link websites in its results, and recommends allowing it in a site's robots.txt file. Google says there are no additional requirements for AI Overviews or AI Mode beyond being indexed and eligible to appear in Search with a snippet.
For any page about you, then, the practical checks are whether these crawlers can reach it, whether Google has indexed it, and whether it says clearly, in text, who you are.
Ask them yourself
Try a few prompts in each assistant:
- Who is [your name]?
- Who is [your name], [your role] at [your organisation]?
- What is [your name] known for?
If the assistant offers a temporary chat, use it, because assistants that remember your previous chats may tailor the answer to you. Note what each one says and, where it shows them, which sources it cites. Those cited pages are the ones shaping the answer.
Why the answers go wrong
Too little to go on. If only a LinkedIn headline and a company page mention you, the assistant pads the gap with generalities, or declines to answer.
Stale pages. An old team page or a three-year-old conference listing may be the clearest statement of your career the system can find, so you appear in a job you left long ago.
Two people merged. Where people share a name, assistants sometimes blend their careers into one answer. Our guide to sharing a name online covers this in more detail.
Invented details. In 2024 the privacy group noyb filed a complaint against OpenAI with the Austrian data protection authority after ChatGPT repeatedly gave a public figure the wrong date of birth, a fact that wasn't published anywhere online. According to the complaint, OpenAI said the only way to stop the wrong date appearing would be to block all information about him.
What you can change
Publish the facts in plain text
Your current role, your organisation, what you work on, notable past roles, where you're based. Put them on a page you control, written out in sentences. Assistants quote and summarise text, and a clear statement such as "She has led the firm's Middle East practice since 2021" is easy for them to use correctly.
Keep the page reachable
If you run your own site, check that robots.txt doesn't block the search crawlers above. Some hosting and security services block AI crawlers by default, so it's worth checking even if you never changed the setting yourself.
Update or retire old pages
Ask former employers to remove or update your profile, and refresh any speaker bios you can edit.
Earn mentions elsewhere
Assistants give weight to what independent sources say. An interview on a trade publication's site or a profile from an industry body counts for more than another page you wrote yourself.
Make your own pages worth citing
In the research paper that coined the term generative engine optimization, published at the KDD 2024 conference, researchers found that adding citations, quotations and statistics to source pages could raise their visibility in AI-generated answers by up to 40% in their tests. For a personal profile, the equivalents are concrete figures from your work, your own words in quotation marks, and links to the sources that back up your claims.
Limits
You can't tell an assistant what to say. Pages written to instruct AI systems, or stuffed with claims about yourself, are more likely to be ignored than repeated, and Google's spam policies now include "attempting to manipulate generative AI responses in Google Search". Nor can you see inside the training data. If an old fact is baked into a model, it may take a newer version of that model to lose it.
Correcting a wrong answer
Most assistants have a feedback button on each answer. It's worth using, though don't expect an immediate change. If ChatGPT shows inaccurate personal information about you, OpenAI has a privacy request process for asking that it stop appearing in responses. It applies under privacy laws such as the GDPR and is open whether or not you have an account. The lasting fix is still to publish the correct information somewhere the live search will find it.
Intervou profiles and AI search
An Intervou profile gives assistants something specific to read: a page about one person, with your role stated in text and a transcript of you answering questions about your work. The page is open to search and AI crawlers, and links out to any website or profiles you give us.
Related guides
- How to rank for your own nameWhat decides the results for a name search, and how to put a page you control at the top.
- When someone else has your nameHow search engines choose between people with the same name, and how to make sure you can be found.
- What is an executive profile?How a profile differs from a bio, a CV and LinkedIn, what a good one contains, and where to publish it.