ChatGPT: Visual Ad Format, Expanded Measurement, and Brand Safety

OpenAI strengthens the advertising infrastructure of ChatGPT. Beyond a visual format being tested during image generation, the company highlights brand safety guardrails, expands its measurement integrations, and shares initial results from partners on performance.

Placement Guardrails and Third-Party Pilots for Brand Safety

OpenAI is introducing advertising policies and guardrails designed to protect users and maintain trust. Advertising spaces are subject to analysis aimed at excluding environments perceived as emotionally vulnerable, sensitive, or inappropriate. The identification and handling of policy violations rely on automated systems, human oversight, and ongoing monitoring. To meet the placement requirements specific to certain brands' policies, negative keyword lists are made available to eligible advertisers. Experiments to assess brand relevance are conducted in partnership with DoubleVerify and Integral Ad Science. These initiatives allow independent partners to evaluate the application of brand safety standards in a controlled testing environment, without access to private conversations. OpenAI aims to provide more transparency regarding these protections while preserving privacy, asserting that ChatGPT's responses remain independent and that users maintain control over their experience.

Expanded Measurement and Initial Results Reported by Partners

OpenAI is expanding its tools and measurement ecosystem for ChatGPT ads by combining internal solutions and partnerships. Integrations with Hightouch, Tealium, and LiveRamp enable the routing of conversion data to the platform. The company relies on web and app attribution partners for conversion, reporting, and click attribution APIs, including AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge, and Tenjin. Advanced measurement and full-funnel partnerships are announced with Fospha, Measured, and INCRMNTAL. OpenAI is also collaborating with Haus, Measured, and WorkMagic on geo-based incrementality experiments, noting that these efforts are in their early stages. Partners are reporting initial results: DV Rockerbox indicates an attributed acquisition cost for WeightWatchers that is 15.3% lower than a mixed paid search benchmark; WorkMagic reports a statistically significant increase for Dose, with 67% of incremental purchases coming from new customers; Triple Whale states that 93% of visitors to Portland Leather from the ads were new. OpenAI thus highlights solid performance through various measurement approaches.

A Visual Format Tested During Image Generation

OpenAI is introducing a visual ad format designed to help users discover products and services and envision their uses. This format, tested during image generation, is presented as clearly labeled and distinct from the image being created. OpenAI clarifies that, as with its other formats, advertising does not influence ChatGPT's responses. The company claims to provide advertisers with new ways to tell their stories through images.

Weekly Reach and Strategic Role of Advertising in ChatGPT

OpenAI states that ChatGPT reaches 1.2 billion people each week. According to the company, advertising helps bring the benefits of AI to a larger number of users, and this broad access is central to its mission and a strength for its business. Advertising allows for serving a market larger than just subscriptions and offers brands a way to reach people in the exploration, evaluation, and decision-making phases. OpenAI positions these updates within an ambition to help users discover, compare, and choose products and services within ChatGPT, and plans to continue evolving its advertising platform with new formats and capabilities to create opportunities for businesses of all sizes.