For the past few months there’s been speculation about whether ads would eventually come to ChatGPT. That question now has a clear answer: they’re coming.
OpenAI has started testing advertising inside ChatGPT, initially with a limited group of users in the United States. The tests focus on people using the free version of ChatGPT and the lower-cost “Go” plan, while paid tiers like Plus, Pro, Business, and Enterprise remain ad-free.
For now, the ads themselves are relatively simple. They appear in clearly labeled boxes at the bottom of responses, separate from the AI’s answer, and only when they’re relevant to the conversation.
If you ask ChatGPT about planning a trip to New York, you might see a sponsored hotel link below the response. Ask about recipes, and you might see an ad for grocery delivery or a meal kit service.
In other words, OpenAI is deliberately avoiding the most controversial approach: embedding ads directly inside the AI’s answer.
That decision matters more than it might seem.
The tightrope OpenAI is trying to walk
AI assistants run on trust. People ask ChatGPT everything from “What laptop should I buy?” to “How do I negotiate my salary?” to “Help me understand this legal contract.” If users ever start to suspect that the answers themselves are influenced by advertisers, the product becomes much less valuable.
OpenAI knows this. The company has repeatedly emphasized that ads do not influence the answers ChatGPT generates, and that advertisers have no way to change or shape responses.
That’s why the early design of ChatGPT ads is intentionally conservative:
- ads are clearly labeled
- ads are visually separated from answers
- ads appear only when contextually relevant
- advertisers never see individual user conversations
Instead, advertisers receive only aggregated performance data like impressions and clicks.
In other words, OpenAI is trying to build something that looks less like social media advertising and more like contextual search advertising. But even if the implementation is careful, the implications are huge.
ChatGPT is quietly becoming a discovery engine
When people talk about ChatGPT, they usually talk about productivity, writing emails, summarizing documents & generating code. But something else has been happening quietly over the past two years, people have started using AI to discover things.
Not in the traditional “search results” sense. In the “conversation” sense.
Instead of typing: best project management software
people ask: What’s the best project management tool for a remote startup of 10 people?
Instead of: running shoes
they ask: I’m training for my first half marathon, what shoes should I buy?
The difference is subtle but important. Search gives you links. AI gives you recommendations. And recommendations are the most powerful form of marketing that exists.
The moment of intent is changing
What makes ChatGPT advertising potentially powerful isn’t the format, it’s the timing. Ads appear exactly when someone is asking for help solving a problem. This is closer to search advertising than social advertising, but with an extra layer of context. The AI understands not just the keywords, but the full situation a user is describing.
If someone says: “I’m planning a three day trip to Paris with my partner and want boutique hotel recommendations” the intent is far clearer than any keyword could capture.
This is the promise of AI advertising: ads tied to real intent instead of simple keyword matching. Whether it actually delivers on that promise remains to be seen.
Why ads were inevitable
There’s also a more practical reason this shift was always coming. ChatGPT is incredibly expensive to run. Training and operating large AI models requires massive computing infrastructure. Meanwhile, hundreds of millions of people use ChatGPT every week, most of them for free.
That creates a simple economic problem. Subscriptions alone are unlikely to support AI at global scale. Advertising, on the other hand, has funded most of the internet for the past twenty years. Seen through that lens, ads inside ChatGPT are not surprising at all. They’re the most obvious way to support a free AI product used by hundreds of millions of people.
The real question for businesses
Right now, these ads are only in testing. They’re limited to certain users, certain regions, and a small group of advertisers reportedly participating in early pilots. But the direction of travel is clear.
At some point, conversational AI will become a major discovery channel, and when that happens, the question for businesses will change from: “How do we rank on Google?” to something more interesting like “How do we become the brand the AI recommends?”
Advertising will likely be one piece of that puzzle. But it probably won’t be the whole story. Because even with ads, AI systems still rely heavily on the information that exists across the web, articles, documentation, product reviews, forums, and educational content.
The companies that consistently show up in that information ecosystem are the ones AI systems learn about. Which means the businesses that teach the internet the most clearly may ultimately become the ones AI recommends the most often.
This is only the beginning
Right now, ChatGPT ads are cautious, limited, and relatively simple. A sponsored box below an answer which is clearly labeled, easy to dismiss, but that’s how every advertising platform starts.
- Google Ads looked simple in 2001.
- Facebook Ads looked simple in 2012.
- TikTok Ads looked simple in 2019.
The interesting changes come later.
New formats, new targeting models, new ways to connect discovery, recommendation, and purchasing.
The testing happening now isn’t the final form of AI advertising, it’s the starting point. For businesses paying attention, it’s an early signal that the next major marketing channel might not be a feed or a search page.
It might be a conversation.