Two of the biggest AI platforms in the world both decided, within a few months of each other, that the free ride was over. Google now sells ads inside AI Mode's conversational answers. OpenAI sells them inside ChatGPT too. Neither company treats the other's approach as the model to copy, and the difference between the two isn't cosmetic — it changes what actually gets optimized when a paid budget moves into this space.
What Google actually shipped
At Google Marketing Live in May 2026, Google introduced Conversational Discovery ads and Highlighted Answers, both powered by Gemini and both placed directly inside AI Mode's generated response rather than beside it. Gemini writes a fresh explainer for each query, evaluating the advertiser's product data and synthesizing why it fits what the person asked, displayed alongside the advertiser's own creative. The ad still carries a "Sponsored" label, but the surrounding context — the part that reads like independent advice — is AI-generated too. According to MediaPost's reporting, this shipped roughly a month before OpenAI's own version of the same idea, timed to AI Mode crossing a billion monthly active users globally.
What OpenAI shipped, and why it's a deliberately different bet
OpenAI took the more cautious route. Testing began on February 9, 2026, limited to logged-in Free and Go tier users in the US, with Plus, Pro, Business, Enterprise and Edu subscribers kept entirely ad-free. The ads appear below a ChatGPT response, not inside it, visually separated and clearly labeled — and OpenAI has been explicit that the placement doesn't touch the actual answer, stating directly that ads don't influence what ChatGPT tells a user. A self-serve OpenAI Ads Manager followed on May 5, 2026, with CPC and CPM bidding and no minimum spend, and the program had expanded to nine countries within a few months.
The tension neither company has fully resolved
Google's approach has already drawn pointed criticism from within the industry it depends on. Sonata Insights founder Debra Aho Williamson flagged that the explainer copy running beside a Google AI Mode ad reads as AI-generated rather than advertiser-written, and pointed to the blending of organic and sponsored content in the same breath as the actual problem, not the presence of ads themselves. OpenAI's more separated format is partly a hedge against exactly that criticism, at the cost of putting sponsored content somewhere users are more likely to scroll past entirely.
What this means for where budget should actually go
The mechanics differ, but what wins in both systems is closer to shared than either platform's marketing suggests. Google's Gemini-written explainer draws on the same structured product data that already determines whether an AI system cites a brand organically in the first place — the same clean feed that helps GEO work also becomes the raw material Gemini uses to write the ad copy around a paid placement. Accurate pricing, real specs and current stock status aren't just a technical SEO checkbox anymore; on Google's side specifically, they're an input the paid campaign gets graded on too.
Budgets built around static keyword bidding and pre-written ad copy are optimizing for a mechanism that's already partly gone on one of the two biggest platforms in the category. The practical move isn't picking a side between Google's blended approach and OpenAI's separated one. It's making sure the underlying paid ads strategy and the product data behind it are solid enough that whichever format a person actually sees, the ad it produces is accurate rather than embarrassing.