Are AI platforms building advertising’s next walled gardens?

First came the question of what an advertisement inside an AI assistant should look like. Then came another: what happens when the platform selling the ad is also building products that compete with the advertisers buying it?

OpenAI has already offered the first glimpse of that tension. As image and voice capabilities have moved inside ChatGPT, the company has restricted advertising by standalone rivals in those categories, turning what appears at first to be an advertising policy into a broader question of platform competition. And, increasingly, platform control.

Also Read: What does an ad look like inside AI? ChatGPT and Google are starting to answer

The First Shot: How a quiet ad policy change signals a new AI platform war

But the larger story is not about one OpenAI policy. It’s about what comes next.

The third question is perhaps the most uncomfortable of them all: who gets to advertise there at all?

As conversational AI platforms move deeper into discovery, recommendations, advertising and eventually commerce, they could begin bringing several parts of the consumer journey under one roof. The assistant may understand what the consumer wants, decide which information to surface, introduce a commercial message, direct the user towards a product and potentially measure what happens next.

The walls are not fully built yet. But the ingredients of a new kind of advertising walled garden are beginning to appear.

Old idea, new walls

Walled gardens are hardly new to advertising. Redseer estimates that closed ecosystems such as Google, Meta and Amazon already capture 70–80% of global ad spend. In India, search and social (dominated by Google and Meta for the most part) accounted for 64% of digital advertising revenues in 2025, according to FICCI-EY.

Conversational AI is now beginning to build an advertising business on top of another potentially proprietary asset: intent. OpenAI says ChatGPT Ads reached a $1 billion annualised revenue run rate in under 200 days, while reported internal projections put its 2026 advertising revenue at around $2.4–2.5 billion.

https://www.exchange4media.com/digital-news/ai-made-the-ad-who-owns-it-indias-copyright-office-just-weighed-in-157931.html

The scale is no longer niche. OneLittleWeb’s June 2026 analysis, using traffic estimates from Ahrefs and Semrush across 10,171 websites, ranked ChatGPT as the world’s fifth most-visited website with 5.32 billion visits, although Google still drew 98.19 billion — 18.5 times as many.

Harjiv Singh, Founder and CEO of CambrianEdge.ai, an artificial intelligence platform, describes this as a “currency change, not a format change”.

Search and social, he noted, have spent years inferring what consumers want from what they click, watch or search for. In a conversation, the consumer may simply tell the system what they are comparing, what problem they need solved and what decision they are trying to make.

“That’s a different currency than reach or impressions, and it shifts planning from targeting audiences to targeting moments,” Singh said.

That could make the conversation itself one of the most valuable proprietary assets an AI platform owns.

For Nikhil Kumar, digital marketing industry leader, the risk lies in how many layers of the advertising relationship could ultimately sit with the same platform.

“There is a risk of a new kind of walled garden emerging, with the platform controlling the intent signal, inventory and measurement,” he said.

The difference may lie in the nature of that first layer: intent.

“Search gave advertisers keywords; social gave them audiences; conversational AI can potentially give them context — what a consumer wants, their constraints and how close they are to a decision,” Kumar said, arguing that different depths of intent could eventually command different prices.

Shifting intentions

Ali Zaidi, Senior VP-Media at Tonic Worldwide, points to the shift from knowing what somebody searched for to potentially understanding why they are searching, which alternatives they are considering, what constraints they face and how close they may be to a decision.

For marketers, he believes that could create an entirely new intelligence layer built around recurring questions, anxieties, consideration criteria and decision triggers.

Two advertisers might ultimately have access to exactly the same inventory, Zaidi said, but the one that better understands the conversational demand underlying it could make that inventory considerably more productive.

The complication is that the platform will usually understand that conversation far better than the advertiser buying against it.

Advertisers neither need nor should receive users’ private conversations. But that creates a familiar digital-advertising problem in a potentially more pronounced form: the company with the richest signal may also define the targeting logic, sell the inventory and report the results.

Kumar argues that advertisers will consequently need transparency around how intent is classified and inventory is valued, independent verification of delivery and measurement, and evidence that conversational advertising generates genuinely incremental outcomes.

“Advertisers don’t need access to the underlying conversation — privacy remains non-negotiable — but they do need confidence that the signal being monetised is real and the value being reported is independently validated,” he said.

That is where the emerging AI stack potentially begins to look different even from the walled gardens advertisers already know.

Viren Inaniyan, Co-founder and CEO of Tru Commerce, an infrastructure platform for agentic commerce, argues that the important development is not simply whether platforms prohibit competing advertisers.

“They will protect whatever they now do themselves,” he said, predicting that different companies will erect walls around different parts of their businesses.

Meta, for instance, may derive more value from using conversations as targeting signals elsewhere across its advertising network. Google can commercialise conversational behaviour through Search. Microsoft has its own combination of Copilot, enterprise software and cloud infrastructure.

“The story is not ‘platforms hate competitors’,” Inaniyan said. “It is that the assistant, the recommendation, the ad slot and the checkout are collapsing into one stack. Whoever owns that stack does not need to sell the competing tool a banner.”

The recommendation layer may prove particularly important.

Making decisions

Amardeep Singh, Co-founder and President of Gutenberg, notes that brands are already having to think about how they are understood organically by AI systems before considering paid visibility around them.

His brand's AI visibility work begins with ensuring that a brand is accurately understood and sufficiently authoritative to be cited inside an AI answer before money is put behind a sponsored placement next to it.

That creates an unusual convergence. The same environment can potentially mediate whether a brand is organically recommended, whether it can pay for commercial visibility, how that commercial message is targeted and eventually how its performance is measured.

There are limits to how quickly advertisers are likely to embrace that ecosystem.

In India, Inaniyan describes current spending as “test money, not channel money”. The more immediate question for brands, he said, is whether conversational AI can prove the value of mid-funnel intent around behaviours such as comparing, shortlisting and choosing products.

Major allocations are unlikely to follow until advertisers have “a measurement story finance will accept”.

And trust remains the constraint on the entire model.

Singh argues that the commercial opportunity only exists while consumers continue to believe the underlying AI answer is independent. Amardeep Singh similarly warns that the value of conversational advertising begins to erode the moment users believe the answer itself has been bought.

That may ultimately determine how high the walls can rise.

The first generation of digital walled gardens was built around proprietary audiences, inventory and measurement. Conversational AI could add something still more valuable to that structure: the context surrounding the decision itself.

Whether those platforms become advertising’s next great closed ecosystems will depend on how much of the commercial journey they eventually control, and how much transparency advertisers and consumers demand in return.

The new walls are not here yet. But the building blocks are already arriving.