ChatGPT has launched ads: what it means for Korean brands

ChatGPT ads have launched. As reported, a test opened on February 9, 2026 for users on the US free and Go plans, and in May a self-serve console that lets advertisers run campaigns themselves was unveiled. Expansion to Korea, Japan, the UK, Brazil, and Mexico has only been announced as "coming weeks," with no exact date yet.

You cannot buy a place in the answer itself. You cannot pay to have ChatGPT cite or recommend a brand within the body of its answer. That's because OpenAI designed ads to be separate from answers (OpenAI Help Center). Whether you run ads or not, getting your name into an answer is something you have to earn through content.

Ads attaching to search is the biggest change since Google AdWords in 2000. So everyone asks "when can I buy ads" first, but there's a more urgent question for Korean brands. It's how to secure that answer spot you can't even buy with money. Below, we lay out the launch situation and the response in order.

Where things stand now: the launch situation

ChatGPT ads rolled out in stages over four months. Because the timing, plans, and regions all differ, Korean brands just need to pinpoint "the stage that applies to me right now."

February–April 2026: the US pilot

As far as is known, the first test targets were adult users logged in to the US free plan and the ChatGPT Go plan. Target, Adobe, Williams-Sonoma, and Albertsons were named as the first advertisers. According to Search Engine Land's early reporting, ads appeared as cards on the first screen, right after the answer. Billing started as per-impression (CPM) bidding, then switched to a per-click (CPC) model in April. The recommended bid was reported to be in the $3–5 range.

May 2026: the self-serve console

In May, a console (ads.openai.com) where advertisers build campaigns themselves opened. The most notable change is that the previous $50,000 minimum spend requirement disappeared. The barrier to entry has clearly come down. There is one condition.

Self-serve sign-up is currently only available to US businesses. Expansion to the five markets including Korea has only been announced at the "coming weeks" level mentioned earlier, and no Korea launch timeline has emerged. The procedures for setting up campaigns, ad groups, and conversion tracking are published in documentation, but it's still too early to transcribe the detailed menu names or internal metric names as-is.

Korean users can't see them yet

Ads currently appear for logged-in free and Go users in the US, Canada, Australia, and New Zealand (the US also includes non-logged-in users). Ads do not appear on the Plus, Pro, Enterprise, or Edu paid plans in any region, and users under 18 are also excluded.

The Korean market is still at a pre-ad stage. On the other hand, a Korean brand running ads into the US market is possible even now, as long as the conditions are met. Treat ad display in Korea and ad campaigns in the US as separate decisions from the start.

Why "Ads Are Separated From Answers" Is the Crux

Ads and answers are physically separated. Almost everything follows from this. It means OpenAI placed the trustworthiness of answers above ad revenue, and thanks to that choice, the value of organic exposure (organic citations) actually rises.

The answer-body spot is not up for bidding

When a user asks a question, a brand's place among ChatGPT's mentions and recommendations in the answer cannot be bought or sold. Ads appear as separate cards beside or below the answer, marked "Sponsored." OpenAI's policy has codified the separation of ads and answers and prohibits advertisers from manipulating responses to insert their own brand into the body of an answer.

For example, suppose someone asks "recommend an ampoule worth buying in the US." There are brands mentioned naturally within the answer's sentences, and there are brands that appear below as ad cards. Users generally read the former as "what the AI picked" and the latter as "paid exposure." And the former spot can't be bought with ad spend; it can only be won through content.

So even when ads open, organic exposure value isn't discounted

Because the two are separate, even when ads come in the value of an answer citation remains intact, and as it becomes scarcer, it actually rises. A brand that captures both spots, ads and answers, takes trust and reach together and begins to pull ahead of competitors who filled only one.

For Korean brands, it works in two directions

ChatGPT ads affect Korean brands in two ways. One is the US-market entry channel that's open now; the other is advance preparation ahead of the Korea launch. Because the operating parties and eligibility conditions differ, it's better to plan them separately from the start.

US market: executable now

A brand with a US entity can run ads now. Eligible categories are consumer goods, travel, education, and local services, while medical, mental health, politics, and finance are blocked by the system. What's needed to register is generally as follows.

• A US entity (Delaware, California, etc.) and an EIN (Employer Identification Number)
• A US business address and a US payment method
• The Persona identity verification process

There are rules for agency structures too: under the terms, only a setup where the advertiser signs up directly first and then invites the agency as a user is allowed. A structure that bundles multiple advertisers under a single agency account is prohibited. PION follows these same rules when running ads on your behalf.

Preparing for the Korea launch: start with content and entities

The timing of Korean ads is still undecided. The problem is what happens when they actually open. As mentioned earlier, the gap between brands already cited in Korean-language answers and those that aren't doesn't close easily no matter how much ad spend you throw at it, because the answer body was never up for bidding in the first place.

So what you should be working on now isn't ad preparation. It's preparation to make it into the answers. Measure how your name is handled in the core questions of your category, and first build the content structure that will get pulled up into answers and consistency in naming. Only when these assets are in place beforehand will ads later be worth their price on top of them.

How PION handles it

PION handles organic exposure (organic AEO) work and ChatGPT ad operations as a single connected flow. It's a structure that passes the signals from ads straight into content priorities. That said, ad operations themselves have clear eligibility limits, and we explain those limits during the first consultation.

Brands with a US entity: we handle ads too

For a brand with a US entity, PION takes on the entire process, from console sign-up to campaign and creative operations, under user permissions on your behalf. The especially valuable data here, in addition to standard ad metrics (impressions, clicks, CTR), is the conversation-based insights OpenAI provides.

As far as is known, once impressions accumulate to a certain level, the topics of anonymized user questions, expressions of hesitation or objection, and even the actual phrasing are captured in reports. From our standpoint, this is raw material for content, as-is. We transfer it straight into FAQ items, into subheadings, into material for rebuttal pieces.

• If we see a cluster of questions with lots of clicks but weak conversion, we reinforce the trust content on the landing page.
• We make question clusters that convert well the top priority for investment in organic content.

Korean entities and regulated categories: ads are blocked

A brand with only a Korean entity can't do self-serve sign-up at all right now, so PION can't run ads on its behalf either. Medical, mental health, politics, and finance are also blocked from ads by system restrictions. In these areas, the only entry route is the single track of organic exposure.

Instead, so you can switch over the moment Korean ads open, we run consulting on establishing a US entity and advance content-asset building beforehand on a separate track. Because if you start building assets after ads have opened, by then it's already too late.

Frequently asked questions

I've heard ChatGPT ads have launched. Is that true?

It's true. As reported, OpenAI began an ad test on February 9, 2026 targeting logged-in adult users on the US free and ChatGPT Go plans, and Target, Adobe, Williams-Sonoma, and Albertsons were announced as the first advertisers. In May, a self-serve console where advertisers run campaigns themselves opened, and the previous $50,000 minimum spend requirement also disappeared.

When will ChatGPT ads launch in Korea?

No exact timeline has been announced yet. OpenAI has only announced expansion to Korea, Japan, the UK, Brazil, and Mexico as "coming weeks," and ads currently do not appear to Korean users. Self-serve sign-up for Korean businesses is also not possible right now; only US businesses can register.

What does ChatGPT Ads mean for our brand?

Because the ad area and the answer area are separated, the value of organic exposure cited in the answer body stays intact and actually grows. Since the answer-body spot can't be bought with ads, even after ads open, the gap between brands cited in the answers to core questions and those that aren't is hard to fill with ad spend. A brand with a US entity can run ads now, while a Korean-market brand should secure answer-exposure assets before launch.

Can I get our brand into the answer body with ChatGPT ads?

No. OpenAI has codified the separation of the ad and answer areas as policy, and ads appear as separate cards with a "Sponsored" label. An answer-body citation is a spot you can secure only through organic AEO work, and because ad operations and content work complement each other, the effect is greatest when you run them together.

What should Korean brands prepare now?

The first thing to do is measure how your brand is handled in AI answers to the core questions of your category. If you record which questions your name appears in, what phrasing it's cited with, and what position it ranks in across ChatGPT, Perplexity, and Gemini, the gaps you need to fill before the ad launch will emerge as data. A brand with a US entity can, alongside this work, run US-market ad tests and use the conversation insights that come out of them as content material.