A Brand Is Now Partly What AI Says About It.

Type a company name into ChatGPT and add four words:

Should I trust them?

In a few seconds, years of brand strategy, press releases, founder interviews, customer reviews, YouTube videos, news coverage and possibly one furious complaint about parking can collapse into a neat paragraph.

For communications teams, this is a slightly rude development. You can spend six months debating whether the word trusted feels too corporate, then a machine writes its own brand summary in six seconds without attending a single workshop.

And that summary may be the first impression.

For years, the standard line was that a brand is not what a company says about itself. It is what other people say about it.

That was already inconvenient.

Now there is another participant.

A brand is increasingly also what AI says about it.

Your new first impression may be machine-written

This is not some distant behavioural shift involving only developers, teenagers and the colleague who insists on putting AI into every meeting agenda.

Pew Research Center reported in June 2026 that 42% of U.S. adults use AI chatbots to search for information. Pew Research Center

That matters because information search is reputation formation.

Before someone buys from a company, applies for a job there, invests in it, recommends it, distrusts it or decides that its £14 face cream probably contains the secret of eternal youth, they research.

Traditional search mostly gave people ingredients: a company website, reviews, articles, videos, forums and social posts.

The user assembled the opinion.

Generative AI can increasingly assemble it for them.

That is a very different product.

Search gives you sources.

AI can give you a conclusion.

Google used to rank the sources. Now it can write between them.

The mechanics make this more interesting.

Google's 2026 guidance says its generative AI features remain rooted in its core Search ranking and quality systems. It describes using retrieval-augmented generation, where current webpages from Google's Search index can be retrieved to help ground AI responses. Google Search Central

Google also says AI Overviews and AI Mode can use query fan-out, issuing multiple related searches across subtopics and data sources before developing a response. Google Search Central

So when someone asks:

Is this company reliable?

the machine may effectively turn one simple question into several smaller ones.

What does the company say about itself? What do other sources say about it? What information appears around its products? Which sources seem useful? What else does the question imply?

The customer sees one prompt.

Behind it, the system can perform something closer to research.

This is why AI is not simply another search engine with nicer sentences.

It is becoming an interpretation layer.

And interpretation has always been where reputation gets interesting.

This is bigger than GEO

Naturally, marketing has already produced an acronym.

Generative Engine Optimization, or GEO, describes attempts to improve visibility inside AI-generated answers. The term has an academic basis: researchers introduced GEO as a formal framework in work published at ACM KDD in 2024. Princeton University

Useful concept.

Dangerous temptation.

Because the moment an industry discovers a new acronym, someone will sell a twelve-step framework for dominating it before lunch.

Google's own guidance is considerably less theatrical. Its current advice says conventional SEO fundamentals remain relevant to generative search and explicitly states that there are no special technical requirements for appearing in AI Overviews or AI Mode. Google also recommends useful, original, non-commodity content rather than mass-producing pages for every imaginable AI query. Google Search Central

In other words, GEO is not SEO wearing futuristic glasses.

The bigger change is that visibility is increasingly influenced by an information environment that extends well beyond the corporate website.

OpenAI says any public website can potentially appear in ChatGPT search, provided its search crawler is not blocked, and ChatGPT search can cite external sources used to construct responses. OpenAI

So the emerging reputation system includes the obvious things:

Your website. Search results. News coverage. Reviews. Interviews. Videos. Third-party mentions. Industry commentary. Public information about your products and executives.

Individually, companies tend to file these under different departments.

Machines have not received the org chart.

The machine sees one reputation system

This is where the organizational problem becomes interesting.

SEO may sit with one agency. PR with another. Social has its own calendar. Customer service handles reviews. Brand is refining the messaging architecture. Someone in IT is still trying to work out who owns the YouTube password.

Internally, these are separate activities.

Externally, they form one information footprint.

Research from Ahrefs makes the relationship worth watching. Its study examined 75,000 brands across ChatGPT, Google AI Mode and Google AI Overviews. YouTube mentions had the strongest correlation with AI visibility at roughly 0.737, while branded mentions across the web also showed relatively strong correlations of around 0.66 to 0.71. Traditional measures such as content volume and several backlink metrics correlated much more weakly. Ahrefs

The usual warning matters here: correlation is not causation.

Publishing 400 YouTube videos containing your company name is not a strategy. It is evidence that you should probably be stopped.

The useful finding is broader.

Brands that exist prominently across the information environment also tend to appear prominently inside AI answers.

Whether those mentions directly cause AI visibility or simply reflect underlying brand strength is still an open question. But either explanation creates the same communications consequence.

Your reputation cannot be managed channel by channel while assuming the machine will politely respect departmental boundaries.

It won't.

It sees the combined picture.

Being cited is not the same as being remembered

There is another useful complication.

Your website can help an AI answer without your brand receiving much credit for it.

Semrush analysed thousands of domain appearances across ChatGPT, Google AI Overviews, Gemini and Google AI Mode in 2026. It found that 61.7% were “ghost citations”: the AI cited a website as a source without naming the brand in the actual answer. Semrush

That creates a distinction marketers will increasingly need to understand.

Citation: the machine used your information.

Mention: the person saw your name.

Those are not the same result.

Your beautifully researched article might help construct the answer while another company's name gets the starring role.

Congratulations. Your content has become an unpaid research assistant.

This is why AI visibility cannot be measured only by referral traffic or citations. Brands also need to understand whether they are actually being associated with the subjects, attributes and expertise they want to own.

AI can be wrong and still shape perception

There is, of course, an elephant sitting politely inside the chatbot.

AI can be wrong.

OpenAI warns directly that search results and citations can sometimes be incomplete, outdated or incorrect and advises users to check authoritative sources when accuracy matters. OpenAI

That limitation does not make AI reputation less important.

It arguably makes it more important.

An inaccurate result halfway down page four of Google is a nuisance.

An inaccurate statement confidently synthesized into the opening paragraph of an AI answer is a communications problem.

Brands cannot control every answer. Nor should they try to scrub legitimate criticism from the internet and call it reputation strategy.

Sometimes the machine says something negative because people have been saying something negative.

There is no schema markup for "please forget the scandal."

The practical goal is less dramatic: make accurate information easy to discover, important claims easy to verify, expertise visible beyond owned channels and factual inconsistencies easier to resolve.

The communications audit needs a new column

Most established organizations already monitor some version of their reputation.

They look at Google results, media coverage, reviews, social sentiment and competitor activity.

The audit now needs another column:

What does AI say about us?

Not once. Repeatedly, across relevant questions.

What does ChatGPT associate with the brand? What does Google AI Mode say? Which sources appear? Which competitors are recommended? What strengths keep recurring? What criticisms keep recurring?

Where is the company absent from conversations it believes it owns?

Where is its content cited but its brand invisible?

Where is information plainly wrong?

These are not purely SEO questions.

They are PR questions, brand questions, content questions, reputation questions and, occasionally, questions for whoever has finally recovered the YouTube password.

The strategic mistake would be handing the whole subject to one discipline simply because the word "optimization" appeared in the acronym.

Reputation is becoming machine-readable

For decades, companies built brand systems mainly for people.

Visual identities needed to be recognizable. Messages needed to be memorable. Websites needed to persuade. PR needed to build credibility. Search needed to make the right information discoverable.

Another requirement is now appearing alongside them.

The reputation also needs to be legible to machines.

That does not mean writing robotic FAQs for ChatGPT or producing 900 nearly identical articles called "Best Enterprise Software Platform for Left-Handed Procurement Managers."

It means creating a coherent public record.

Clear factual information on owned channels. Consistency in names, products, people and claims. Original material worth citing. Credible third-party coverage. Useful video. Structured information where appropriate. Reviews that receive attention rather than neglect. Expertise demonstrated in places other than the corporate About page.

None of this is particularly futuristic.

That is precisely the point.

Most of the building blocks are old. What has changed is the interpreter sitting above them.

AI is not simply another communications channel. It is becoming an interpreter of every communications channel.

Brands spent decades learning how to speak.

Then they learned how to be found.

Now they also need to understand how they are being summarized.

Because the next potential customer may never read the About page that took six months to approve.

They may simply ask the machine.