ChatGPT Ads Is Now in 89 Countries. Most Brands Will Enter the New Ones Wrong.

On October 21, 2026, OpenAI switched on ChatGPT Ads in 26 more countries. The platform now sells inventory in 89 markets, and OpenAI says more than 100,000 advertisers have run a campaign since the US pilot began in February. Fourteen of the new countries are in Latin America, including Argentina, Chile, Colombia and Peru. The rest are scattered across Europe, the Middle East, Africa and Asia, with Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh and Pakistan named in the announcement.

For a brand that already runs ChatGPT ads, this is the third expansion in nine weeks, after 31 European markets in August and seven in Southeast Asia and Taiwan in September. The temptation is to treat each one the same way: open the campaign's country list, tick the new boxes, let OpenAI translate the copy, and move on.

Whitebox has been monitoring the questions, the answers and the ads on this surface since April. What the data shows about new markets is specific and a little uncomfortable: the advertisers change almost completely from country to country, the fields are thinner but concentrate harder, and the brands that got there first mostly got there in the wrong language. This post is about how to enter a new ChatGPT Ads market without repeating that.

Key Takeaways

How fast this happened

Eight months ago ChatGPT had no ads. The pace since then matters, because every market OpenAI adds resets the competitive clock for anyone selling there.

Market counts and the 100,000 figure are from OpenAI's announcements of August 18, September 23 and October 21, 2026. The 60,000 active-advertiser estimate is from The Information, citing New Street Research, as reported by AdExchanger. OpenAI's "tens of thousands" and "100,000+" are not directly comparable: the first described advertisers running at the time, the second counts everyone who has ever run a campaign.

Two other details from the same week are worth holding onto. OpenAI told AdExchanger it is "building out measurement alongside the expansion", which is a polite way of saying the measurement is not there yet in the new markets. And Digitimes reported on October 20 that OpenAI is cutting ad prices and adding campaign tools to pull in more advertisers. Cheaper inventory in a new country is an invitation to spend before you can measure.

What a new ChatGPT Ads market looks like from the inside

Whitebox's OpenAI Ads market report covered 153,514 sponsored placements across 7,171 questions and 3,352 advertisers between April and August 2026, in the US, Canada, Australia, Japan, South Korea and Brazil. The non-US observations were small compared with the US, and we said so in the report. But they are the only published look at what happens on this surface when a market is young, and three patterns held everywhere we looked.

1. The competitors change, not just the language

We ran one 31-question enterprise security set, translated, in three markets. The United States drew 76 distinct advertisers. Japan drew 34 and South Korea drew 29. Not one advertiser appeared in both the US and either Asian market. We expected the opposite: global software brands advertising everywhere while the organic answers localised. In practice the organic answers overlapped more than the ads did. For US versus Japan, 21.0% of the brands named organically appeared in both markets, and 0% of advertisers did.

Language is not the explanation, because the same thing happens between English-speaking countries. In a devtools category run in the US and Australia, 38.5% of organically named brands appeared in both markets. Only 5.4% of advertisers did.

The practical consequence for the 26 new countries: whatever you know about your competitors on ChatGPT in your home market tells you almost nothing about who you will meet in Colombia, Nigeria or Ukraine. The roster there will be built from local players, regional brands and whichever global advertisers happened to tick the box, and it will be built in the next few weeks.

2. Thinner fields concentrate harder

Fewer advertisers sounds like good news. It is, briefly. In the same enterprise security set the leading advertiser held 21.0% of impressions in the US, 28.5% in Japan and 47.5% in South Korea. Half the advertisers, more than twice the concentration. When the field is thin, the first brand to show up consistently takes a share that is very hard to dislodge later, and on a surface where 84.2% of repeat asks return a different advertiser, consistency is the whole game.

This is the real argument for moving early in a new market, and it is not the one most brands make. The opportunity is not that the inventory is cheap or empty. It is that a defensible position is cheaper to build while the field is thin than to buy back once a local competitor holds half the impressions on your category's questions.

3. The brands that got there first mostly got there wrong

Across the whole sample, 97.1% of the 3,740 ads served against non-Latin-script questions came back in Latin script. In the Japanese categories we monitor, 3.4% of ads were written in Japanese and 3.1% pointed at a Japanese-localised landing page. In Korea both figures were 0.0%. The user asked in Japanese, ChatGPT answered in Japanese, and the sponsored card underneath was in English and sent the click to an English page.

Those advertisers won impressions anyway, because nobody local was competing yet. That will not last. The 26 October markets include at least a dozen languages with non-Latin scripts or non-English defaults, and the first local advertiser who writes a Spanish ad for a Spanish question in Bogotá, pointed at a Spanish page, has a relevance advantage that no bid will overcome.

The convenient path into a new market, and why it is a trap

OpenAI knows the localisation problem exists, and its answer arrives on October 29. From that date, creative enhancements are on by default for self-serve campaigns: OpenAI can translate your copy, rewrite headlines and descriptions, resize images and send the click to a different page on your site, built from the conversation context. For a brand expanding into 26 countries at once, that is an extremely convenient toggle. Tick the countries, leave the setting on, and OpenAI handles the Spanish, the Ukrainian and the Bengali for you.

The problem is what you cannot see afterwards. Reporting stays at campaign, ad group and ad level with no split between the original and the enhanced versions, and OpenAI says you cannot preview every variation. So in a market you have never advertised in, against competitors you have never seen, in a language your team may not read, you will know what you spent and not what ran. If the translated headline misstates a price, or the swapped destination lands on an English page, the first signal will be a conversion rate you cannot explain. We covered the seven checks to run before the switch flips in our creative enhancements checklist; the short version for new markets is to keep the toggle off for any campaign where you cannot read the output.

How to enter one of the 26 new markets properly

None of this requires a big budget. It requires doing four things in the right order, before the campaign goes live rather than after the first report.

See the questions people ask in that market, in their language

Your home-market prompt list is a starting point, not a target list. The questions that carry demand in a new country are phrased differently, mention different competitors, and often name different products. Before building an ad group, you need the actual conversations ChatGPT is having about your category in that country, the organic answer it gives, and which brands it names. If ChatGPT already recommends a local brand on the question you are about to buy, your ad is going to be arguing with the assistant.

Check who is already there, and what they are saying

In a market that is weeks old, the competitor list is short and readable. Find out how often any ad appears on your category's questions in that country, which advertisers hold the share, and whether their creative is local or imported. An incumbent running English copy on Spanish questions is a competitor you can beat on relevance alone. An incumbent with a local page and local copy is one you need to out-position, and you should know which before you write a headline.

Write the ad in the language of the question, pointed at a page in that language

This is the single largest gap in the data and the cheapest to close. Generate the creative from the local question and the local organic answer, in the local language, and send the click to a landing page the user can read. If you do not have that page yet, launch the campaign after you build it, not before. A relevant ad to a generic English homepage is still the second-most common failure in our sample: 52.0% of ads with a URL landed on a bare homepage or category page.

Measure against a control from day one

OpenAI's own measurement partners, from AppsFlyer and Adjust to Haus and Measured, are listed for the US. In a market that opened last week, assume none of that is wired up. Put UTM parameters on every destination before launch, confirm the conversion event fires from the local page, and run the local creative as a structured test against a control rather than reading a single campaign's first fortnight, which on a rotating single-slot surface is mostly noise.

A different way into a new market

OpenAI's route into the 26 new countries is geo-targeting plus automatic translation: tick the box, and the platform adapts your home-market ad on its own. Whitebox was built on the opposite premise. The adaptation is the same, but it happens where you can see it, before it runs, against the actual conversations in that market. Three things make that a different way in rather than the same thing with more dashboards.

You see the market before you spend in it

Whitebox tracks your category's questions per country and per language as separate scopes, and watches the organic answer and every ad served against them. For a market you are about to enter, that means you already know the ad rate on your questions, which advertisers hold the share, what their headlines say and whether they bothered to localise, and which brands ChatGPT itself recommends there. That is the competitor roster you will actually face, not the one from your home market, and it is visible before the campaign exists.

Local creative you approve, not a translation you cannot see

Whitebox generates ad groups, context hints and creative from the real questions in that market, in the question's language, pointed at the matching local page, and maps every ad to the conversations it would match so you can see the fit before launch. Every one of those is a draft you read and approve. OpenAI's translation runs after the fact and never shows you the version that served. For a language your team does not read, the difference between a draft you approved and a rewrite you cannot inspect is the difference between a launch and a gamble.

Tests with a control, and tracking that survives the click

Whitebox runs local creative as structured A/B tests with a control, flags untracked destination URLs and missing conversion integrations before the campaign goes live, and applies UTM tracking across the account in one step. Rules for bids and budgets run with your thresholds, and every change sits in a review queue with its reasoning until someone approves it. In a market where OpenAI is still "building out measurement", that is the measurement.

Underneath all of it, Whitebox is a Generative Engine Optimization platform first. It tracks whether ChatGPT, Gemini, Perplexity and Google AI Overviews name and recommend your brand in each market, which sources they rely on, and how that changes week to week. Entering a new country with ads while the organic answer there recommends somebody else is the expensive way to do it. Seeing both in one view is how you decide whether to buy the slot, fix the answer above it, or both.

Frequently Asked Questions

Which countries did OpenAI add to ChatGPT Ads on October 21, 2026?

OpenAI added 26 countries, bringing ChatGPT Ads to 89 markets. Fourteen of the new countries are in Latin America, including Argentina, Chile, Colombia and Peru. The rest are spread across Europe, the Middle East, Africa and Asia, and OpenAI named Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh and Pakistan. OpenAI has not published the complete list of 26.

How many advertisers are on ChatGPT Ads?

OpenAI says more than 100,000 advertisers have run campaigns since the platform launched in February 2026. That is a cumulative figure. The Information, citing New Street Research, put active advertisers at about 60,000 in early October 2026.

Who sees ads in ChatGPT in the new countries?

The same rule applies in every market: ads are shown only to users on the Free and Go plans. Plus, Pro and Enterprise subscriptions stay ad-free. Advertisers can buy through OpenAI technology partners or self-serve through Ads Manager.

Does an ad written for the US work in a new ChatGPT Ads market?

Usually not well. In Whitebox monitoring of the OpenAI Ads surface, 97.1% of ads served against non-Latin-script questions came back in Latin script, and in Japan only 3.4% of observed ads were written in Japanese. The advertiser roster also changes by country: no advertiser appeared in both the US and Japan for an identical enterprise security question set, and only 5.4% of advertisers overlapped between the US and Australia in a devtools category, even though both are English-speaking.

Will OpenAI translate my ads automatically for new markets?

From October 29, 2026, OpenAI turns on creative enhancements by default for self-serve campaigns. The setting lets OpenAI translate your copy, rewrite headlines and descriptions, and swap the destination URL for another page on your site, without showing you which version ran. You can turn it off in Ads Manager under Manage settings, at the account level or per campaign.

Whitebox figures in this post are from the OpenAI Ads Market Report 2026, based on 153,514 sponsored placements observed between April and August 2026 and published on this blog. They describe the monitored sample, not the whole market, and the cross-market findings rest on two categories. Nothing here measures cost, clicks or conversions. Expansion facts are from OpenAI's announcements; measurement partner names are from OpenAI's October 5 visual ads announcement as reported by TechCrunch; the price-cut report is from Digitimes, October 20, 2026.

See who is already advertising on your category's questions in the 26 new markets, and what ChatGPT says there before your ad appears.

Book Now