$1.14 trillion. That is the projected size of the global advertising market in 2025, growing at 8.8% annually. AI chatbot companies want a piece of it, and they have started making moves.
OpenAI announced advertising trials for ChatGPT on January 16, 2026, with live testing rolling out in February. The early implementation looks familiar: sponsored blocks at the bottom of responses, separated by a line and labeled "Sponsored." One demo showed a meal kit service ad appearing after a cooking-related query. The system matches ads to conversation topics and past interactions.
But as a new analysis from Dr. Felix Simon at the Reuters Institute for the Study of Journalism lays out, banner-style ads at the bottom of a chat are just one end of a much wider spectrum.
The Three Tiers of Chatbot Advertising
Simon maps chatbot advertising across a transparency scale:
Tier 1 - Labeled ads. What OpenAI is testing now. Distinct sponsored blocks, clearly marked, appearing alongside responses. This is the safest option for users and the easiest for regulators to accept.
Tier 2 - Sponsored sources. When a chatbot cites references, some could be paid placements with small "Ad" labels that most users would scroll past. Think of how Google's search ads became nearly indistinguishable from organic results over the years.
Tier 3 - Invisible placement. The chatbot recommends a specific brand or shows a product in a generated image without disclosing the commercial relationship. Simon notes this would face regulatory barriers in many countries, but the incentive is obvious: it is the most valuable ad format precisely because users would not recognize it as advertising.
The Netflix Playbook
The pattern here is not new. Netflix swore it would never run ads, then launched an ad-supported tier that pulled in over $1.5 billion in revenue. Google's own founders wrote in their original Stanford paper that "advertising-funded search engines will be inherently biased towards the advertisers." They built one anyway.
AI companies face the same math. Development costs are enormous, and most users do not pay for subscriptions. Perplexity tried advertising, then walked it back entirely. But the financial gravity is strong. When you have hundreds of millions of users engaging in detailed, personal conversations, the advertising value per impression dwarfs a traditional search result.
Who Gets Hurt
The winners are predictable: incumbent ad platforms like Google, Meta, and Amazon already have the advertiser relationships, and OpenAI gets a new revenue stream.
The losers are worth paying attention to. News publishers, already hit hard by the shift from search traffic to AI-generated answers, face another revenue squeeze. If chatbots absorb the ad dollars that used to flow through publisher websites, the economics of online journalism get even worse.
Then there are the users. The "conversational intimacy" of chatbots - the fact that people ask them personal health questions, financial advice, and relationship problems - makes advertising feel fundamentally different from a banner on a webpage. A chatbot that subtly steers you toward a paid recommendation while you think you are getting objective advice is a different kind of manipulation than a sidebar ad you can ignore.
Simon puts it plainly: "Systems optimised for advertising revenue could easily end up prioritising commercial outcomes over user welfare when the two conflict."
The technical safeguards matter here. Clear labeling, opt-out mechanisms, restrictions on political ads, and protections for minors are all on the governance wishlist. Whether any of them survive contact with quarterly earnings targets is the real question.