ChatGPT prepares a US visual-ad test alongside generated images
OpenAI is preparing a limited US test of visual ads displayed while ChatGPT users generate images. The trial is planned for later October with an initial advertiser group. OpenAI says advertisements will be labeled and separate from the created image. The format accompanies new purchase-measurement partnerships and brand-suitability evaluations, whose partners the company says will not access private conversations.
Artificial Intelligence··Midday
The first visual-ad trial is planned for the United States
OpenAI, the developer of ChatGPT, announced on October 5 that it was preparing visual advertisements for the chatbot’s image-generation experience. A first test is scheduled for later October with a limited group of advertisers in the United States. The placements appear while a user generates an image. This is a planned trial of a new format, with the initial geography and advertiser group specified in the announcement.[1], [2]
OpenAI says an advertisement remains separate from the image being created and carries a clear label. The company also says advertising does not shape ChatGPT’s answers. These assurances describe the provider’s intended product behavior. Visual placements can illustrate a product’s use, suggest a product idea or depict an experience connected with a service.[2]
Purchase measurement accompanies the new format
OpenAI is expanding tools for advertisers to measure clicks and purchases alongside the visual format. Hightouch, Tealium and LiveRamp are named for conversion-data connections. Click-attribution partnerships include AppsFlyer, Adjust, Branch, Triple Whale and Kochava. These integrations concern the actions associated with an advertising campaign, including purchases or other conversions. The company also lists Fospha, Measured and INCRMNTAL for broader campaign measurement across the purchasing journey.[1]
Regional experiments are another part of the measurement program. OpenAI is working with Haus, Measured and WorkMagic on comparisons between geographic areas to assess advertising effects. ChatGPT’s existing advertising program began earlier in the year for free and lower-cost plans, then extended to India in August. The newly announced image-related format has its own limited US trial timetable.[1]
Brand-suitability partners are kept outside private conversations
OpenAI is preparing brand-suitability evaluations with DoubleVerify and Integral Ad Science. The company says these partners will not receive users’ private conversations. It describes safeguards against advertisements in emotionally vulnerable or otherwise unsuitable conversations, combining automated checks with human review. The announcement offers the planned evaluations and protections, without an independently measured failure rate for the safeguards.[1], [2]
The announcement also adds negative-phrase controls for qualifying advertisers. These controls are included among the tools OpenAI describes for managing its advertising placements.[1]
