Evidence at a glance
Expansion Is About More Than Geography
On October 21, 2026, OpenAI announced that ChatGPT Ads would enter 26 additional countries and territories, bringing its stated reach to 89. The service had been running for a little over eight months, and OpenAI said more than 100,000 advertisers had run campaigns on the platform. The announcement also put weekly ChatGPT usage above 1.2 billion. The first figure is cumulative advertiser use, not the number currently active; the second is OpenAI's reported usage figure, not a statistic accompanied by an independent audit in the announcement.
The expansion matters for more than its geography. OpenAI is placing a commercial touchpoint in conversations where people explore, compare, and narrow down options. Search ads often wait for an explicit query, while ChatGPT Ads may reach people before their needs have fully taken shape. For product and technical leaders, that means turning the process of working out what one wants into potential ad inventory—and bringing the timing of recommendations, and the trust they require, into product design.
Matching Happens During the Decision Process
OpenAI's help center describes a matching system that goes beyond pairing an ad with a keyword. It can take into account the current conversation's context and intent, the ad's landing page, headline and copy, advertiser-provided context prompts, and targeting settings. If a user enables ad personalization, some signals from other ChatGPT experiences may also be used. When several ads fit, ranking considers both relevance and advertiser bids.
The mechanism reflects how conversational needs evolve. Someone training for a first marathon might ask about meals for race week, then add dietary preferences and the time available for cooking, before a grocery delivery service becomes relevant. OpenAI says ads can appear below a response, are labeled and kept separate from the answer, and cannot be used by advertisers to determine or change that answer. This limits direct advertiser control over responses, but it does not make ad timing neutral: a product surfaced before a user has decided may still shape what they consider next.
Privacy Promises Are Not a Substitute for Trust
OpenAI says advertisers cannot see users' chat histories, memories, or personal information, and that it does not sell customer data. Ad performance information is provided in aggregated, non-identifying form. These commitments describe what advertisers can access, but they do not mean ads are unrelated to conversations. Matching is explicitly based in part on the context and intent of the current conversation, and enabled personalization may draw on some signals from other ChatGPT experiences.
The product question is therefore not only whether data is handed to advertisers. It is also whether users can understand why an ad appeared, manage personalization, and distinguish advertising from an answer in the actual interface. The available material sets out OpenAI's claims about privacy and presentation, but does not fully specify user controls for personalization or establish how users experience the labeling in practice. For a conversational product that depends on repeat use, these are not peripheral compliance details; they are conditions for the ad model to endure.
89 Countries Do Not Mean 89 Self-Serve Markets
Fourteen of the newly added markets are in Latin America, with Argentina, Chile, Colombia, and Peru named as examples. Other markets mentioned include Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh, and Pakistan. OpenAI says advertisers can access the service through technology partners or Ads Manager. The announcement does not name all 26 additions, so its examples should not be mistaken for a complete list.
More importantly, ad availability and self-serve eligibility are different things. A country being covered does not automatically mean every local advertiser can open an account and launch a campaign directly; some markets require a partner. For international teams, 89 describes the platform's stated reach, not a ready-to-buy list of self-service markets. Entry plans still need to verify local channels and campaign requirements market by market.
A Pilot Should Test Acceptance as Well as Conversion
OpenAI links advertising revenue to supporting free and low-cost access. Ads appear only on Free and Go plans; Plus, Pro, and Enterprise remain ad-free. That makes advertising another way to support service access alongside subscriptions, but its commercial value still depends on advertisers receiving effective reach over time. The announcement's figure of more than 100,000 advertisers is cumulative. AdExchanger reported a New Street Research estimate of roughly 60,000 active advertisers in early October. The figures use different dates and definitions, so they are not necessarily contradictory—and the cumulative count should not stand in for active demand.
Asmita Dubey, L'Oréal Group's chief digital and marketing officer, said the company was testing early with OpenAI and exploring experiences spanning discovery, evaluation, and agentic commerce. The announcement did not disclose specific campaign placements or results. Teams considering a pilot should start with categories where needs emerge through clarification, confirm the local route to market, and track business outcomes alongside user reactions to ad timing, labeling, and relevance. The available evidence does not establish an effectiveness advantage for conversational ads, or provide a complete account of market availability and maturity. Expanded reach creates an opportunity to test, not a guarantee of results.