AI Business

OpenAI expands ChatGPT ads pilot to five more markets while promising answer independence

OpenAI says ChatGPT Ads has launched in the UK, Mexico, Brazil, Japan and South Korea, extending a carefully limited test beyond the United States.

Published Updated
OpenAIChatGPTAdvertisingAI business

OpenAI has expanded its ChatGPT advertising test to the United Kingdom, Mexico, Brazil, Japan and South Korea, its first named enlargement of the program beyond the United States. The update, published August 11, frames the move as a continuing pilot rather than a broad launch. It matters because the company is trying to add a new commercial layer to a service used for work, learning and personal questions without making the answer itself feel bought.

The company says ads are intended to support wider access to ChatGPT’s free and lower-cost tiers. In the original U.S. test, logged-in adults on Free and Go were eligible; Plus, Pro, Business, Enterprise and Education accounts were excluded. That distinction remains central to the product design. Paid plans retain an ad-free experience, while the consumer tiers become the setting in which OpenAI can learn whether advertising can help fund capacity without changing the basic bargain users expect from the assistant.

OpenAI’s public line is that advertisements will not influence answers. Sponsored material is meant to be visibly labeled and separated from the organic response, even when the system judges an ad relevant to the topic of a conversation. In practice, that separation will be watched closely: unlike search results or social feeds, a chat can feel like a single, continuous exchange. The challenge is not only to make a label visible, but to keep the recommendation logic and the answer-generation logic credibly distinct.

Privacy is another constraint the company has emphasized. Advertisers are not to receive chat transcripts, chat history, memories or personal details; they receive aggregate performance information such as views and clicks. OpenAI says it also excludes accounts known or predicted to belong to people under 18 and does not allow ads beside sensitive or regulated topics including health, mental health and politics. Those boundaries address obvious risks, but their practical operation across different markets will be part of the test’s significance.

The expansion also arrives as the economics of consumer AI remain unsettled. Running a widely available assistant requires substantial computing infrastructure, and providers have been experimenting with subscriptions, enterprise plans, developer services and partnerships to cover those costs. Advertising offers another route, but one with a different set of incentives. It could make lower-cost access more durable, while also creating pressure to prove that commercial targeting does not erode the usefulness or perceived neutrality that brought people to the product.

OpenAI says users can dismiss an ad, provide feedback, see why an ad was shown, delete ad data and manage personalization. The company has said early pilot signals included low dismissal rates and no observed effect on its consumer-trust measures, though it has not published a detailed methodology or market-by-market data. That leaves independent evaluation difficult for now and makes the next phase of user feedback more consequential than a conventional product expansion.

For the five new markets, the immediate change is an availability update, not a claim that the final ad format has been settled. OpenAI says it plans to keep learning and to expand to more markets during the year. The test will therefore be judged on more than revenue: users and regulators will be looking for clear separation from answers, meaningful choice, credible safeguards for younger and sensitive audiences, and evidence that a conversational interface can introduce advertising without weakening the reasons people rely on it. It will also show whether the same principles are understandable in different languages and consumer contexts, where expectations about marketing, privacy and platform responsibility are not identical. Public scrutiny is likely to concentrate on the everyday moment of use, when a sponsored suggestion appears beside a request that may have begun as a private search for help.