Advertising has started appearing inside conversational AI products. In Korea, OpenAI began showing visually separated ads below responses for free and Go tier users on June 19.
Where things stand
Google and Naver are also expanding tailored ads against shopping-intent queries in AI search and briefing surfaces. With OpenAI, turning on ad personalization lets the assistant serve relevant ads based on signals collected during use, while Plus, Pro, Business, Enterprise and Edu subscribers are excluded from ads entirely.
Why now
OpenAI framed the move as necessary to sustain infrastructure and investment. Competition is part of the picture: ChatGPT's share in Korea sits in the 40% range, below the global average of 46.4%, while Gemini has climbed to roughly 30% — raising the pressure to monetize.
The real issue is format
On a search results page, paid and organic results are visually distinct. A conversational answer is one continuous passage. If ads drift toward native placement inside that passage, users lose the ability to separate a recommendation from a purchase. That matters more than it did with banners, because the answer is already doing the judging for the reader.
Disclosure requirements are tightening in adjacent areas — see DV360 Structured Data Files v10.1 Adds AI Labels and DOOH for how labeling obligations are spreading through ad tooling.
What marketers should prepare
First, treat presence in an AI answer as a different metric than a search click: inclusion comes before the click. Second, run paid placement and organic citation as two separate tracks — buying your way into an answer does not replace being cited for credibility. Third, note that ad-free paid tiers mean your reachable audience is now segmented by subscription level.
The cost of getting ad format wrong is measurable. "Connected Content": 80% Say a Bad Ad Is Worse Than No Ad at All puts a number on it.