OpenAI to test visual ads in ChatGPT during image generation, adds measurement tools

OpenAI is expanding ChatGPT advertising beyond text placements. A new visual format will appear while ChatGPT is working, notably during image generation, which can take several seconds. That waiting time becomes ad inventory. The test covers Free and Go tiers in the US, starts later this month with an initial group of advertisers, and comes with new attribution and measurement partnerships.

Digiday frames this as OpenAI fixing advertiser frustrations that date back to the February ad launch: too few formats, limited inventory and weak measurement. Ads shown during generation solve the inventory problem neatly, because they fill dead time without inserting sponsored content into the answer itself. That distinction matters for trust, since paid placements inside answers would be far more contentious.

Scale is the reason this matters. Sources cite about 1.2 billion weekly active users, which would make ChatGPT one of the largest ad surfaces outside Google and Meta. That puts OpenAI into more direct competition with both companies for brand budgets.

User reaction is predictably negative. Hacker News threads express frustration with ads intruding into generative interfaces, even as advertisers welcome the measurement tools. Another live thread asks whether Plus subscribers will cancel as Chat and Work merge, a sign of broader product-change fatigue. Watch for ad load, whether paid tiers stay ad-free, and early CPM and attribution data from launch advertisers.