OpenAI Begins Internal Testing of Advertising Placements in ChatGPT

The prevailing fear about artificial intelligence chatbots has long been that they would eventually weave paid product placements directly into their answers, compromising the integrity of the information. The reality of OpenAI's advertising rollout is far more conventional. As the company expands its internal testing and public deployment of ChatGPT Ads, the evidence shows these placements are strictly separated from the generative text, appearing as distinct, labeled cards at the bottom of the screen rather than native product integrations.[1]

This week, OpenAI significantly expanded its advertising infrastructure, introducing a suite of tools designed to turn ChatGPT into a mature performance marketing channel. The company is currently testing multi-advertiser placements, which allow several brands to compete for space within a single sponsored unit via a second-price auction model.[3]

Alongside the auction mechanics, OpenAI is testing a multi-product carousel format specifically designed for retail and e-commerce feeds. This allows advertisers to display a range of goods in one placement, moving away from the single-product cards that characterized the initial ad tests earlier this year.[4]

The testing coincides with a massive geographic expansion. OpenAI is rolling out ChatGPT Ads to users in India, Brazil, and Mexico, opening up inventory to over 100 million active users across those regions. The ads are exclusively targeted at users on the Free and lower-cost Go tiers; subscribers to the Plus, Pro, Business, and Enterprise plans remain entirely ad-free.[1]

OpenAI is pitching these placements to agencies as a way to reach consumers at "high-context moments when decisions are beginning to take shape." However, early performance data suggests the platform is still finding its footing. Media buyers report that while impression volumes are high, click-through rates are hovering around 0.6 percent—substantially lower than the 2 percent average typically seen in traditional search advertising.[1][5]

To close that performance gap, OpenAI is deploying conversion-optimized bidding, allowing advertisers to pay per click rather than per thousand impressions. The company has also integrated with third-party measurement tools like Triple Whale and Hightouch, and introduced Automatic Advanced Matching to improve attribution tracking for performance marketers.[4][6]

A critical component of the testing phase involves user privacy. OpenAI executives have stated categorically that advertisers do not receive access to users' conversations, chat histories, or personal details. The ad matching is handled internally by OpenAI's systems, which use the context of the current prompt to select a relevant sponsored card without passing that raw text to the bidding brands.[1]

As the platform matures, media agencies are already calculating where the ad dollars will come from. Industry analysts expect ChatGPT Ads to primarily cannibalize existing search and performance-social budgets. Because the chatbot requires a user to initiate a specific query, it functions as a demand-capture tool rather than a demand-generation medium.[2]

Consequently, mass media budgets—such as television, outdoor, and cinema—are expected to remain largely untouched by the shift. Those channels excel at creating broad awareness from scratch, a function that conversational AI interfaces are not currently designed to replicate.[2]

The ongoing tests of carousel formats and conversion bidding signal that OpenAI is aggressively courting performance marketers who demand measurable outcomes. As the platform scales its self-serve Ads Manager across new international markets, the true test will be whether these high-context placements can eventually match the conversion rates of traditional search engines.[3][4]