ChatGPT starts selling images: a visual ad format tests inside image generation, with early attribution data from three partners

ChatGPT gains a visual ad format that presents product inspiration, product usage or the experiences a product enables as images, so people can learn about a brand and an offering while viewing a picture. Earlier ad formats did not carry a product story through imagery; the new format gives advertisers a way to tell that story visually. The format is first tested during image generation in ChatGPT for users on the Free and Go plans, with ads clearly labeled and kept separate from the image being created.

Whether a conversation is appropriate for advertising decides placement, and emotionally vulnerable, sensitive or otherwise unsuitable contexts are kept out. Brand controls extend from general policies to narrow placement requirements specific to a brand's own policies, with Negative Phrases available to qualifying advertisers. OpenAI says its placement guardrails assess whether a conversation is appropriate for advertising, backed by automated review, human oversight and ongoing monitoring to identify and address violations.

On the measurement side, data paths and the partner ecosystem expand so advertisers can send conversion data from systems they already use into ChatGPT Ads. Advertisers previously lacked a way to measure how ChatGPT Ads contribute to their business through the tools they rely on; they can now connect through those tools. Hightouch, Tealium and LiveRamp provide conversion-data integrations, while AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge and Tenjin provide Conversions API, reporting and click attribution.

Incrementality measurement is framed as the next step toward understanding advertising's business impact, currently explored through geo-based experiments. Attribution data records which conversions are credited to ads, while incrementality measurement tries to answer whether ads produced conversions that would not otherwise have happened. OpenAI describes this work as still in its early stages and says it is partnering with Haus, Measured and WorkMagic to explore geo-based experiments.

The demonstrated scope is a visual ad format tested during image generation in ChatGPT for Free and Go users, starting later this month in the US with an initial group of advertisers, alongside measurement expansions covering conversion-data integrations, attribution partners and geo-experiment exploration. The intended readers are engineers and product researchers following AI-native ad formats and measurement infrastructure.

OpenAI says its incrementality work is still in its early stages; brand suitability evaluation is a pilot with DoubleVerify and Integral Ad Science intended to assess how brand safety standards are applied in a controlled testing environment without accessing private user conversations.