Imagine if there were a company
that knew exactly what you’re currently looking for,
understood why you started searching in the first place,
and might even grasp your budget, preferences, constraints, and real hesitations,
would it then understand better than Google or Facebook:
Which ads are most likely to make you take action right now?
This question used to be a hypothetical one.
Now it is becoming a real business model for ChatGPT.
On September 8, 2026,
OpenAI CFO Sarah Friar spoke at the Goldman Sachs Communacopia + Technology Conference about ChatGPT’s advertising potential, using a very direct metaphor:
It’s like Google and Meta had a child.
This phrase is noteworthy not because OpenAI suddenly thought of selling ads.
ChatGPT ads had already begun.
The real importance is:
OpenAI is now more openly recognizing
why conversational AI could become a unique advertising channel, different from search engines and social platforms.
The Most Valuable Aspect of Google: Knowing “What You Want Right Now”
First, consider Google.
When you search for:
“Tainan hotel”
“Tokyo flight tickets”
“Best 50-inch TV”
“Accounting software for small companies”
These keywords have high commercial value.
Because you’re not just interested in the topic.
You’re likely ready to take action.
This is what’s called intent.
Intent.
A person searching for “Osaka hotel” today
is very different from someone who liked Japanese travel photos six months ago.
The business value is completely different.
The former might book a room this week.
So one of Google Ads’ most important capabilities has been:
Knowing what you want at the exact moment you’re searching for it.
What Meta Values Most: Knowing “Who You Are”
Facebook and Instagram work differently.
You may not be searching for sneakers right now.
But the platform knows:
Which accounts you follow.
What videos you watch.
Which content you linger on.
Which ads you previously clicked.
What topics tend to interest you.
These signals gradually create a longer-term context,
which means:
What kind of person you usually are.
Meta doesn’t have to wait for you to search.
It can predict what you might be interested in.
ChatGPT Can Potentially Combine Both
This is the real meaning behind Sarah Friar’s “Google + Meta” comment.
In ChatGPT,
people rarely just input two or three keywords.
You might say:
“I plan to take my parents to Japan for 5 days in October. They’re in their 70s, can’t walk too much, budget around 60,000, and want a place with convenient transportation.”
That’s no longer just “Japan hotel.”
It includes:
Time.
Travel companions.
Age.
Budget.
Mobility limitations.
Transportation preference.
Travel purpose.
You might even add:
“Breakfast isn’t important, but it’s best if there’s a convenience store nearby.”
This information would normally be scattered across different websites, searches, apps, and behavior data.
But in conversational AI,
the user lays out the needs clearly themselves.
This is why ChatGPT’s advertising model is worth watching.
OpenAI Officially Uses “Conversational Context”
This is not just Sarah Friar’s imagination.
OpenAI’s own ChatGPT Ads documentation makes this very clear.
When selecting ads, the system considers:
The current conversation context and intent.
That is:
What you’re talking about right now,
and what you seem to want to accomplish.
If the user enables ad personalization,
the system may also reference some signals from the wider ChatGPT usage experience.
Advertisers can also provide context hints,
telling the system which types of conversations, topics, or situations their products suit best.
This is very different from simply buying ads on:
“Japan hotel”
“Running shoes”
“Accounting software”
based purely on keywords.
How Much More Information Does a Conversation Provide Compared to a Keyword?
Imagine two people both search for:
“Recommended laptops.”
A says:
“Recommended laptops.”
B tells ChatGPT:
“I want to buy a laptop for my university daughter, mainly for reports and simple video editing, hoping it lasts 4 years, budget around 30,000, no gaming, and not too heavy.”
On the surface,
both are searching the same topic.
But for an ad system,
B has already offered many conditions:
Student.
Budget.
Weight.
Expected usage time.
Video editing.
No gaming.
This is one of the biggest business differences between conversational AI and traditional search.
It’s not necessarily about knowing more people,
but rather:
Knowing more deeply about a specific decision.
The Most Interesting Part: Sam Altman Was Originally Very Reluctant
Going back to 2024.
At that time, OpenAI hadn’t formally turned ChatGPT ads into the platform it is now.
When asked about ads,
Sam Altman was quite reserved.
He said,
Ads would be a “last resort.”
He specifically noted:
“The combination of ads and AI makes me especially uneasy.”
Looking back now,
the contrast is striking.
Because less than two years later,
ChatGPT ads not only appeared,
they have become a very big business.
In Under 200 Days, Annualized Revenue Hits $1 Billion
On August 31, 2026,
OpenAI officially announced:
ChatGPT Ads launched less than 200 days ago,
and the annual run rate has reached:
$1 billion.
And it’s not just a few large brands experimenting.
OpenAI states
> that tens of thousands of advertisers are already using the platform.Ads are also expanding to more countries.
Ads Manager allows small and medium businesses to create campaigns themselves,
and the system now supports:
CPM,
CPC,
conversion tracking,
product feeds,
location targeting,
and custom audiences.
Sound familiar?
That’s because these are the same foundational features built by Google Ads and Meta Ads over many years.
ChatGPT is now entering this market.
But OpenAI Clearly Wants More Than Just Banner Ads Below Answers
Currently, ChatGPT ads are still quite ordinary.
They appear below the answers.
Clearly marked as sponsored.
With title,
description,
image,
and link.
In style,
You can think of them as:
Sponsored items shown after AI delivers its response.
However, the key point Sarah Friar made on September 8 is:
This isn’t the end form.
What excites her is:
OpenAI hasn’t yet launched
a truly “native AI” ad format.
This statement is worth remembering.
Because if in the future ChatGPT ads are just Google Search Ads with a different UI,
then there’s no revolution.
The real market changer will be:
Ads themselves start understanding conversations.
For Example, Ads Might No Longer Just Ask, “Do You Want to Buy?”
Today, many ads are basically:
You searched for shoes,
I show you shoes.
You searched for hotels,
I show you hotels.
But conversational AI might understand:
Why haven’t you bought yet?
For example, you might say:
“This camera looks great, but I rarely take photos and worry it might go to waste.”
Your real issue
is not which camera is best.
It’s:
You’re not sure if you really need it.
If the ad system understands this level,
brands might shift from saying:
“Now 10% off.”
to
“Try renting it for seven days first.”
This is where context truly adds value.
Not by making ads follow you more,
but by making them closer to
the decision you are currently making.
But This Also Touches a Very Sensitive Line
Because ChatGPT differs greatly from search engines.
People say in AI many things they wouldn’t say in a search box.
Work struggles.
Family arrangements.
Financial concerns.
Relationship issues.
Health questions.
Life decisions.
So when “conversation context” starts holding ad value,
the first inevitable question is:
Can advertisers see what I said to ChatGPT?
OpenAI’s official answer right now is:
No.
Ads don’t affect ChatGPT’s answers.
Advertisers don’t get users’ private conversations.
OpenAI also promises not to sell user data to advertisers.
So it’s important to differentiate two things here.
“The system uses conversation context to decide which ads are relevant”
versus
“Giving advertisers access to your full conversations.”
They are not the same.
OpenAI’s current public policy implements the former, not the latter.
That’s Why “Ads Should Not Influence Answers” Is So Important
Imagine one day you ask:
“Which laptop suits me best?”
ChatGPT answers A.
And below, you see an ad for brand B.
As long as these two are independent,
users can understand:
The top is AI’s suggestion.
The bottom is an ad.
But if someday,
advertisers paying more could make AI recommend their products more strongly in the answer,
ChatGPT’s value would instantly change.
Users would no longer know:
Is it advice,
or an ad?
So OpenAI now emphasizes repeatedly:
Ads are separate from answers.
This separation may seem like just UI design,
but it’s actually the most crucial trust line for ChatGPT’s ad model.
Also, Not Everyone Sees Ads
Under OpenAI’s current rules,
ChatGPT ads mainly appear on Free and Go plans.
Plus, Pro, and Business plans don’t show ads.
Accounts identified as under 18 aren’t shown ads either.
OpenAI also sets sensitive context limits,
avoiding ads near personal, high-risk, or emotionally vulnerable conversations.
All these restrictions say the same thing:
ChatGPT’s ad issue
isn’t just “put more ads and earn more.”
It has to answer:
Under what circumstances are people willing to accept commercial messages?
The True Competition May Shift from “Who Gets the Most Traffic” to “Who Understands Intent Best”
In the past 20 years,
the two main players in online ads have been Google and Meta.
Google captures massive amounts of search intent.
Meta captures extensive social, interest, and behavior context.
Now ChatGPT introduces a third element:
Conversation.
Dialogue.
People inside it aren’t just clicking.
Nor only searching.
They’re explaining their problems.
This could give AI platforms a brand-new commercial signal:
Not “what you’ve looked at,”
Not “what similar people bought,”
but
what you are currently trying to achieve.
This is why Sarah Friar’s “Google + Meta” analogy is so notable.
What she really sees
isn’t just two ad formats combined,
but two crucial ad assets overlapping:
Intent + Context.
Yet ChatGPT’s Most Valuable Asset May Ultimately Be Trust
Google can know what you search for.
Meta can know what you watch.
ChatGPT’s biggest advantage might be that you’re willing to tell it directly:
What you really want to do.
This is a very precious relationship.
But it’s also very fragile.
If one day,
users start to doubt:
“Are your answers today driven by who paid?”
Then no matter how much context ChatGPT has,
the value of that context could quickly vanish.
So OpenAI’s real challenge now
is not how to turn ChatGPT into Google + Meta,
but:
How to generate Google + Meta level ad revenue without becoming an ad machine in users’ eyes.
In 2024, Sam Altman was uneasy about “ads + AI.”
By 2026,
that concern has become the business reality OpenAI must solve.
The next big battlefield for AI advertising
may not be about whose ads are more accurate,
but
when AI already knows what you’re thinking, who can still make you believe it’s helping you first, not selling to you?
Today, let's grow a little with AI.
Learn an AI tip every day.
Save a little time daily.
Enhance your skills step by step.
SasaDaily, growing with you.
Recommended Reading
AI Quick Q&A｜2026/07/13: Does AI Peek Into My Uploaded Files? Which Data You Should Never Upload
