OpenAI is bringing ads into the free version of ChatGPT, starting with a limited rollout in the US. The experiment is still changing, but early testing shows how different chatbot advertising can feel compared with older search ads: the placement sits directly beneath an answer, inside a conversation users may treat as private and personal.
In one rough test of 500 questions on the ChatGPT mobile app, ads appeared often enough to become hard to ignore. About one out of every five questions in a new conversation thread triggered an ad at the bottom of the response. Each ad included a website link as a button, and the ad topic generally followed the topic of the latest prompt.
How the ads show up
The ads seen in the test were not random banners floating outside the product. They were attached to ChatGPT answers, which makes their placement feel closely tied to the exchange itself. OpenAI says it is moving slowly, beginning with a limited set of advertisers and ad formats while it learns from the rollout.
The range of ads was broad. The test surfaced ads for dog food, printers, hotel reservations, productivity software, movie tickets, food delivery apps, fashionable ties, streaming services, corporate credit cards, apartment furniture, cruise vacations, AI coding tools, freelance editors, skin-care articles, business internet plans, handmade gifts, grocery stores, and basketball tickets, among others.
Some examples showed how directly an ad could connect to a question. A prompt about the gig economy produced an Uber ad that read "Your Schedule, Your Earnings." A question about the worst TV show ever led to an ad for Page Six's Hollywood newsletter. A comparison involving Harvard versus Stanford triggered an ad for the University of Minnesota's part-time MBA program.
Travel questions seemed especially likely to produce ads. When the tester asked for help planning a trip to Palm Springs, ChatGPT displayed a Booking.com ad. Clicking the ad opened a search for hotel deals in Palm Springs.
What OpenAI says is happening
OpenAI started testing ads in ChatGPT in February in the US, and the tester first noticed them in a tester account in early March. The company says the ads do not affect the content of ChatGPT answers. It also says a user's full conversation is not shared with companies that pay for ad placements.
That still leaves a meaningful targeting system in place. According to the source, ads can be influenced by the topic of the user's question, past chats, and whatever ChatGPT stores in its memory about that user. That matters because chatbot prompts can reveal immediate needs, plans, preferences, and decisions in a way that feels more intimate than a conventional search box.
OpenAI has framed the move as part of a longer-term effort to keep ChatGPT broadly accessible, not as a response to any rumored IPO later this year. The company also says its early signals are positive. An update on OpenAI's website dated March 26 says it is seeing no impact on consumer trust metrics, low dismissal rates of ads, and improving relevance as it learns from feedback.
After the limited US rollout, OpenAI is expected to expand the ad push to Canada, Australia, and New Zealand.
The business logic is clear
The commercial case is not difficult to understand. Free users are costly for OpenAI, and online search habits are shifting as people ask generative AI tools for information and practical guidance. That creates a new opening for advertising around answers, recommendations, and task planning.
Olivier Toubia, a marketing professor at Columbia Business School who focuses on AI, described the emerging market as connected to money now spent on search ads. His view is that generative AI-focused advertising could become a huge multibillion-dollar market.
The test also showed a familiar digital advertising tactic inside a new interface. When a prompt named a brand such as DoorDash or Netflix, the ad below the answer sometimes came from a direct competitor. Toubia describes that approach as "poaching," a long-running feature of search advertising that may now carry over into large language model advertising.
OpenAI is also hiring around the effort. The company has open roles tied to ad integration, including a "product marketing lead, advertising." Part of that role involves identifying risk areas such as "performance, safety, policy, trust" and driving plans to mitigate them.
Trust is the hard part
The risk is that ads may change how people interpret ChatGPT's answers, even if advertisers do not influence the output. Sam Altman previously voiced discomfort with the idea. At Harvard Business School in 2024, he said, "I hate ads," and called the mix of "ads plus AI" uniquely unsettling.
That concern is sharper in a chatbot because the product is conversational. A sponsored button under an answer can make a user more aware of what they have disclosed, what the system remembers, and why a particular company appeared at that moment.
Stefano Puntoni, a marketing professor at Wharton who researches generative AI, warned that ChatGPT cannot easily reduce the quality of the experience without risking users leaving. Google’s Gemini and Anthropic’s Claude do not currently feature sponsored ad buttons in outputs, though Google recently said it is not ruling it out.
OpenAI's challenge is therefore not simply adding ads. It is adding them without making the answer feel bought, the conversation feel watched, or the product feel less useful. If users begin to doubt the neutrality of recommendations, the core value of a chatbot weakens quickly.
That is why the rollout matters beyond any single ad for hotels, food delivery, or software. ChatGPT ads are an early test of whether generative AI can support advertising inside a personal assistant interface while keeping the confidence that makes people use it in the first place.