OpenAI is preparing to test a more visual form of advertising inside ChatGPT, placing sponsored product or service imagery alongside image-generation sessions. The experiment will begin in the United States later in October with an initial group of advertisers, according to OpenAI’s October 5 announcement and reporting by The Verge.

The format represents a step beyond the sponsored placements OpenAI introduced in February. The Verge reports that those earlier ads displayed a company name, logo and product link beneath a conversation. The new units are designed to show product inspiration, use or related experiences through images, giving advertisers more room to present an offering visually.

OpenAI says the ads will be clearly labeled and kept separate from the image a user is generating. The company also says advertising will not affect ChatGPT’s answers. The Verge reports that visual ads will not appear for subscribers to ChatGPT Plus, Pro or Enterprise plans, preserving an ad-free experience for those tiers.

An abstract image-generation workspace keeps a visual ad card outside the creative output.
The first test is limited to image-generation sessions in the United States with an initial advertiser group.

The initial placement is narrow: OpenAI says visual ads will first be tested during image generation. That context could give advertisers a natural opening when users are already working visually, but it also makes the boundary between generated content and sponsored material especially important. The company’s stated design keeps the ad outside the created image rather than embedding a product into the output.

OpenAI is pairing the new format with a larger measurement push. Its announcement lists integrations with Hightouch, Tealium and LiveRamp for sending conversion data, along with attribution and reporting support involving partners including AppsFlyer, Triple Whale, Adjust, Branch and Kochava. It is also exploring geo-based experiments with Haus, Measured and WorkMagic to estimate whether ads caused additional business results.

The company supplied early partner-reported performance figures, though these should be treated as vendor claims rather than independent evidence. OpenAI said DV Rockerbox found that WeightWatchers’ attributed acquisition cost on ChatGPT Ads was 15.3 percent below its blended paid-search benchmark. It also cited WorkMagic and Triple Whale findings involving Dose and Portland Leather.

A privacy shield and review symbols filter ads around a protected conversation.
OpenAI says placement guardrails and external suitability pilots will help keep ads out of sensitive contexts.

Placement safety is another central part of the rollout. OpenAI says its guardrails assess whether a conversation is suitable for advertising and are intended to keep ads away from emotionally vulnerable, sensitive or otherwise inappropriate contexts. The company describes automated review, human oversight and continuing monitoring, while The Verge notes that these protections determine when a conversation is considered appropriate for an ad.

OpenAI is also developing brand-suitability pilots with DoubleVerify and Integral Ad Science. The company says those partners will evaluate its standards in controlled settings without accessing private user conversations. Qualifying advertisers can also use a Negative Phrases feature for narrower placement restrictions tied to their own policies.

The advertising expansion is strategically significant because OpenAI says ChatGPT reaches 1.2 billion people each week and argues that ads can support access beyond paid subscriptions. The visual test remains limited, and neither The Verge nor OpenAI establishes how users will respond once sponsored imagery appears beside creative work. The experiment will test whether OpenAI can build a larger advertising business while maintaining a credible separation between commercial placement, private conversation and model output.