How ChatGPT decides what to say about you

AI assistants are becoming part of the modern background check.

A few years ago, someone researching you might type your name into Google, open several links, and make their own judgment.

Today, that same person may ask:

• Who is John Smith?
• What is John Smith known for?
• Is John Smith trustworthy?
• Has John Smith ever been involved in a lawsuit or controversy?
• What should I know before doing business with John Smith?

Instead of receiving ten blue links, the user may receive a single paragraph that sounds authoritative and complete.

That changes online reputation management in an important way.

The question is no longer only:

What appears when someone searches your name?

It is also:

What story will an AI system construct from the information it can find about you?

Those two things are connected, but they are not identical.

AI does not have a permanent biography about you

People sometimes imagine that ChatGPT, Gemini, Perplexity, Copilot, Grok, or another AI system has a permanent file containing everything it knows about them.

That is not really how it works.

An AI-generated answer may be influenced by several different sources of information.

Depending on the system and the question, those can include:

• Information learned during the model's training
• Live web search results
• Pages retrieved during the conversation
• News articles
• Company websites
• Professional profiles
• Government and regulatory pages
• Court-related sources
• Wikipedia or other reference sites
• Social profiles
• Interviews
• Videos and transcripts
• Previously indexed content
• Sources specifically cited by the AI system

Some AI systems rely more heavily on live retrieval than others. Some show citations. Some summarize information without clearly showing where every claim originated.

That means your AI reputation can change even when the underlying model itself has not been retrained.

The web around you changes.

New pages are published. Old pages gain or lose visibility. Professional profiles are updated. Search rankings move. News coverage changes. Websites disappear. Better sources become available.

Those changes can affect the information an AI system encounters when someone asks about you.

Why one negative page can have an outsized effect

Suppose someone searches your name and finds:

• Your LinkedIn profile
• Your company biography
• An old lawsuit
• A professional directory
• An old news article about the lawsuit
• A mostly empty personal website

A human researcher may understand that the lawsuit happened years ago and weigh it against your current career.

An AI assistant may not apply the same judgment.

If the lawsuit page is highly detailed, published by a strong website, clearly connected to your full name, and repeated by several other pages, it may become one of the easiest sources for the AI to summarize.

The result can be disproportionate attention to a relatively small part of your life.

This is especially common when there is not much authoritative, current information available about you.

If the web contains thousands of words describing one negative event and only a two-sentence LinkedIn biography explaining everything else you have done, the information environment is badly unbalanced.

The AI did not necessarily decide that the negative event was the most important part of your life.

It may simply have found far more accessible information about that event than about anything else.

Search visibility still matters

Traditional search rankings remain important because many AI systems use web search or retrieval when generating current answers.

A page that is highly visible for your name may therefore influence both:

• What a human researcher sees
• What an AI assistant finds

This is one reason traditional search-result suppression remains relevant in the AI era.

If a negative article dominates your name search, it has more opportunities to be encountered.

If the first page instead contains a strong collection of accurate, current sources, the information available to both people and machines becomes more balanced.

Those positive sources might include:

• Your personal website
• Current company biography
• LinkedIn
• YouTube
• Medium
• Professional associations
• Interviews
• Credible articles
• Industry publications
• Speaking profiles
• Community involvement
• Current news coverage

The objective is not to trick an AI system.

The objective is to give it better information.

Source strength matters too

Ranking is not the only consideration.

AI systems may place more confidence in certain sources because of their apparent authority, specificity, structure, or relevance.

A government page describing a regulatory action may carry more weight than an Instagram post saying you are a great person.

A detailed newspaper article may be easier to use than a thin personal profile.

A well-developed professional biography may be more useful than ten nearly empty social accounts.

That means effective AI reputation management cannot simply involve opening dozens of profiles.

The assets need substance.

They need to clearly explain:

• Who you are
• What you do
• Where you work
• Your professional history
• Your areas of expertise
• Your accomplishments
• Your current activities
• Your relevant organizations
• Your public work
• The correct context around important events

Consistency also matters.

If your LinkedIn profile says one thing, your company biography says another, and your personal website has not been updated in five years, an AI system may have difficulty determining which version is correct.

Why AI answers can be wrong

AI systems can produce incorrect or misleading information for several reasons.

Two problems are especially important in personal reputation management.

### Stale information

The internet has a long memory.

A complaint may have been withdrawn.

A lawsuit may have been dismissed.

A regulatory issue may have been resolved.

A company may have changed ownership.

You may have left a job years ago.

An old article may still rank prominently without reflecting what happened afterward.

AI systems can sometimes repeat that old information without providing sufficient context.

For example:

> John Smith is the CEO of ABC Company.

That statement may have been true five years ago and completely false today.

Or:

> John Smith was sued for fraud.

That may technically describe an allegation in a complaint while omitting the fact that the case was later dismissed.

The information may not be entirely fabricated, but the resulting impression can still be wrong.

### Confusion with another person

This is particularly dangerous for people with common names.

Two people may share:

• The same first and last name
• The same profession
• The same state
• Similar ages
• Similar employers

If the surrounding web data is unclear, an AI system can sometimes blend information from different people.

A doctor in New Jersey may suddenly acquire the disciplinary history of a doctor with the same name in Florida.

An executive may be associated with another person's lawsuit.

An attorney may be credited with another lawyer's case history.

This is why building a clear digital identity matters.

Your website and professional profiles should consistently connect:

• Full name
• Middle initial when useful
• Photograph
• Employer
• Location
• Profession
• Education
• Credentials
• Relevant profile links

The clearer your identity is across multiple credible sources, the easier it becomes for both search engines and AI systems to distinguish you from everyone else.

AI can also fill gaps

Another problem occurs when there simply is not enough information available.

If you have very little online presence, an AI system may attempt to construct an answer from a handful of weak or incomplete sources.

That creates what might be called an information vacuum.

Imagine that the web contains:

• One old negative news story
• An incomplete LinkedIn profile
• A people-search listing
• A court document
• No personal website
• No articles
• No videos
• No current professional biography

The negative information does not need to be overwhelmingly powerful.

It simply has very little competition.

This is why one of the first goals of a reputation-management campaign is often to build a proper foundation.

The solution is usually not to fight the AI answer directly

When someone sees an unfavorable AI answer, the natural response is:

How do I make ChatGPT change this?

Sometimes there may be mechanisms for reporting clearly incorrect information.

But reputation management usually requires looking underneath the answer.

Ask:

• What sources are influencing this?
• Which claims appear repeatedly?
• Which sources are outdated?
• What accurate information is missing?
• Which negative pages dominate the search results?
• Which customer-owned assets are weak or nonexistent?
• Are there credible third-party sources telling the current story?
• Are the AI systems finding inconsistent information?

The AI answer is often the symptom.

The information environment is the underlying problem.

Build better sources

The long-term strategy is to create and strengthen accurate sources that explain who you are today.

That can begin with customer-owned assets such as:

• A personal domain
• A substantive personal website
• A complete LinkedIn profile
• A YouTube channel
• A Medium profile
• Relevant social profiles

Then the strategy can expand into stronger third-party assets:

• Professional directories
• Industry publications
• Interviews
• Press coverage
• Association profiles
• Podcasts
• Conference appearances
• Credible contributed articles

Not every customer needs every platform.

The right mix depends on the strength of the negative results, the person's profession, search volume, existing assets, budget, and the sources already influencing search and AI answers.

Suppression still matters

Suppression is sometimes misunderstood.

It does not mean deleting history from the internet.

It means building stronger and more relevant assets that compete for visibility.

If a negative result currently appears in position two, the campaign may work to create enough credible alternatives that the result eventually moves lower.

That same process can also improve the information available to AI systems.

As stronger pages become more visible, AI assistants may have more current and balanced material available when constructing an answer.

There is no guarantee that a specific AI system will immediately change its response.

But improving the underlying source environment is far more durable than trying to manipulate a single chatbot answer.

You need to monitor more than one AI system

There is no universal AI answer.

ChatGPT may describe you one way.

Google AI may describe you another way.

Perplexity may emphasize a different source.

Gemini may find newer information.

Grok may include information from a different part of the web.

Answers can also change depending on the exact question.

Compare:

• Who is John Smith?
• What controversies are associated with John Smith?
• Is John Smith a reputable financial professional?

Those are three different prompts and can produce three very different reputational pictures.

Modern reputation monitoring therefore needs to consider both:

• Multiple platforms
• Multiple relevant questions

What Mirror.fyi does

Mirror.fyi is designed around this broader reputation environment.

The platform begins by examining what currently appears around your name and identifying the sources most likely to shape how people and AI systems understand you.

It evaluates:

• Positive results
• Negative results
• Neutral results
• Customer-owned assets
• Third-party assets
• Search-result strength
• Reputation difficulty
• Missing foundational assets
• AI-generated summaries
• Potential inconsistencies
• Areas where additional content is needed

From there, Mirror.fyi develops a personalized action plan.

That may include strengthening an existing profile, building a personal website, publishing an article, creating video content, developing stronger third-party assets, or addressing inaccurate information.

The customer executes the work and retains ownership of the resulting assets.

Mirror.fyi continues monitoring the environment and adjusts the strategy as the campaign develops.

The most important principle

You cannot control exactly what an AI system will say about you.

But you can influence the quality of the information it has available.

A thin digital identity leaves machines and people dependent on whatever information happens to be easiest to find.

A strong digital identity creates context.

It establishes current facts.

It distinguishes you from other people with similar names.

It gives search engines more credible pages to rank.

And it gives AI systems better material from which to construct their answers.

That is why AI reputation management ultimately begins in the same place effective online reputation management always has:

Improve the sources, and the story can begin to change.