OpenAI announced this week that it will begin testing advertisements inside the ChatGPT app for some US users — a move that marks a pragmatic turn from CEO Sam Altman's earlier distaste for ads. The banners, OpenAI says, will appear outside of the model's written responses: boxed, labeled and placed above or below answers rather than woven into the assistant's text.
What users will actually see
Starting in the coming weeks, logged-in users of the free ChatGPT tier and the new $8-per-month ChatGPT Go plan can expect to see small, contextual banner ads in the chat window. OpenAI's mock-ups show a tinted block with an image and short copy — for example, travel offers after an itinerary request. Paid subscribers on Plus, Pro, Business and Enterprise tiers will not see these ads.
The company emphasizes limits: ads won't be served to people it determines are under 18, and they won't appear alongside answers about sensitive topics such as health, mental health or politics. OpenAI also says it will not share conversation content with advertisers and that ad content will not influence the assistant's responses.
Fidji Simo, OpenAI's CEO of applications, framed the experiment as part of a broader strategy to diversify revenue while keeping the assistant trustworthy. 'We believe in having a diverse revenue model where ads can play a part in making intelligence more accessible to everyone,' she wrote.
Why OpenAI is testing ads now
Money. The company is facing enormous infrastructure and model costs as it scales. Public reporting and company comments over the past year have made clear that subscriptions alone — only about 5 percent of ChatGPT's roughly 800 million weekly users pay — aren't covering all expenses. OpenAI has signaled it expects big annual capital outlays and, per filings reported in the press, forecast large cash burn even as revenue climbs. That pressure helps explain the shift from Altman's previous rhetorical resistance to advertising in the product he helped make ubiquitous.
Altman himself has been candid about his reservations: he once said he 'kind of hate[s] ads just as an aesthetic choice' and warned in public appearances that a chatbot whose answers could be shaped by commercial pressure would undermine trust. The ad placements OpenAI is testing look designed to answer that critique: separate boxes, clear labels and promises that the model's output remains advertiser-free.
The mechanics and the risks
OpenAI says ads will be shown 'when there’s a relevant sponsored product or service based on your current conversation.' That makes them conversationally contextual, which may improve relevance — and creepiness. Critics worry that even boxed ads will change user behavior, attention patterns and trust. Tech critic Ed Zitron wrote bluntly on social media that he is 'extremely bearish on this ads product,' arguing ads won't move the needle enough against the scale of OpenAI's costs.
There are also privacy and safety vectors to consider. OpenAI promises not to share conversations with advertisers, but contextual ads imply some level of processing tied to the content of a chat. Users and regulators will want clarity about what data is used for ad selection, how long it is retained, and whether any profiling crosses thresholds set by privacy laws.
A broader pattern across AI products
OpenAI is not alone in experimenting with commerce and ad models in conversational experiences. Google and others have been testing ad placements and shopping features inside chat-style interfaces, and OpenAI itself has rolled out shopping-related functions before. This signals a broader industry move to blend conversational assistance with commercial intent — for better or worse. For a sense of how OpenAI is diversifying its product footprint, note its push into other consumer experiences such as Sora on Android, which underscores the company's appetite to expand where its models live and how they make money (OpenAI's Sora lands on Android).
Meanwhile, rival products are folding commerce and bookings into everyday flows — for example, conversational map and travel assistance that can suggest and book experiences through AI-powered interfaces (Google Maps gets a Gemini AI copilot). Those moves make ad-based or affiliate revenue models a natural, if contentious, option.
What this means for users and businesses
- Free users will likely see the biggest change: a small banner after certain answers.
- Paying customers on higher tiers get an ad-free experience, which strengthens the value proposition for subscriptions.
- Advertisers gain a new, conversational placement where relevance feels high — and where the line between helpful suggestion and paid placement will be watched closely.
OpenAI says it will test, learn, and iterate based on feedback. That iterative approach matters: if the ad units feel invasive or if users suspect the assistant's outputs are biased by sponsors, trust — a core asset for a conversational AI — could erode quickly.
Expect debate. Some see ads as an inevitable step to underwrite massive compute and data costs; others see it as a compromise that risks the product's independence. Either way, the experiment will be a high-profile test of whether advertising can coexist with an AI people depend on for judgment, creativity and personal assistance.