ChatGPT tests visual ads during image generation

Waiting for an AI image to generate can feel like watching digital paint dry. OpenAI just found a way to monetize that pause.

The company announced a new visual ad format for ChatGPT, rolling out to test with Free and Go tier users in the US later this month. These image-based ads will appear specifically while users are generating images, with OpenAI promising clear labels and zero impact on the actual AI answers or creative outputs.

With 1.2 billion weekly active users, ChatGPT is already the largest AI-native surface on earth. Now it is building the machinery to rival traditional search giants for ad dollars.

The conversion plumbing goes live

Pretty pictures alone do not convince media buyers. To get performance marketers on board, OpenAI rolled out a massive ecosystem of measurement integrations.

Advertisers can now funnel conversion data into ChatGPT Ads via Hightouch, Tealium, and LiveRamp. For attribution, OpenAI linked arms with industry staples including AppsFlyer, Triple Whale, Adjust, Northbeam, Branch, and DV Rockerbox. Incrementality testing is also in the works with partners like Haus, Measured, and WorkMagic.

The early performance figures will turn heads in growth marketing circles. According to DV Rockerbox, wellness brand WeightWatchers logged a cost per acquisition 15.3% lower than its paid-search benchmarks.

Even more telling is the new customer discovery. Triple Whale reported that 93% of visitors hitting Portland Leather through ChatGPT Ads were entirely new. WorkMagic tracked a similar trend for wellness label Dose, with 67% of incremental sales coming from first-time buyers.

Building safety moats around the prompt

The biggest nightmare for enterprise brands is popping up alongside sensitive prompts or emotional distress. OpenAI says its automated guardrails will assess context to block ads from emotionally vulnerable conversations.

To prove it without exposing user chats, OpenAI is running brand suitability pilots with verification heavyweights DoubleVerify and Integral Ad Science (IAS). Qualified advertisers can also deploy negative phrases to stay away from topics that clash with their brand guidelines.

Why it matters

For digital marketers, ChatGPT is quickly transitioning from an experimental novelty into an essential performance channel. Reaching 1.2 billion users at the exact moment they are researching problems or seeking inspiration is prime real estate. The sky-high share of net-new customers reported in early tests suggests the audience here behaves very differently from traditional searchers, making the waitlist at ads.openai.com worth joining immediately.

For the tech industry, OpenAI just fired a direct shot at Google's and Meta's core cash engines. Subscription revenue was never going to fund the staggering compute bills required to train frontier models. Ads were inevitable, but by embedding rich visuals right into the image generation latency window, OpenAI found a clever way to turn a technical bottleneck into ad revenue.

We think the execution here is technically sharp, but the cultural stakes are massive. Search engines built trust on separating blue links from sponsored slots, and Google spent two decades defending that boundary. If users ever start suspecting that ChatGPT's recommendations are quietly swayed by its growing ad business, that hard-won conversational trust could evaporate overnight.