OpenAI to Test Ads Next to ChatGPT Images in US Starting Late October

OpenAI will begin testing 'visual ads' that display product and service advertisements alongside ChatGPT image generation results, starting with select advertisers in the US in late October 2026. OpenAI announced the plan on October 5 (local time).

What's Changing

• While users generate images, ads related to products or services will appear next to the generation results.
• The ads will be clearly distinguished from user-generated images and marked with a 'Sponsored' label.
• The format focuses on visually conveying usage and experience, not just the product itself.
• OpenAI emphasized, as with its existing ads, that advertising does not influence ChatGPT's responses.

This test is part of a strategy to expand ChatGPT's advertising business, centered on free and lower-tier plans. According to AI Times, ChatGPT has 200 million weekly active users (WAU), and OpenAI plans to expand ads globally going forward.

What Tools Are Available for Advertisers

• Conversion Data IntegrationThrough partnerships with Hightouch, Treasure Data, and LiveRamp, advertisers can now pass conversion data from their existing systems into ChatGPT ads.
• Performance MeasurementOpenAI has partnered with AppsFlyer, Triple Whale, Adjust, DV Lockerbox, Northbeam, Branch, Singular, Kochava, Airbridge, and Tenjin, among others, to enable measurement of clicks, conversions, and performance across web and app.
• Incrementality ReviewOpenAI is exploring regional experiments with Haus, Measured, and WorkMagic to assess the causal impact of ads on actual purchases and other outcomes. While still at an early stage, OpenAI explained that the goal is to capture true incremental effects rather than just click or conversion counts.
• Exposure ManagementPlacement guardrails are applied to prevent ads from appearing in sensitive or emotionally vulnerable conversations unsuitable for advertising, with automated review, human oversight, and continuous monitoring used to check for policy violations.
• Brand SafetyOpenAI offers a 'Negative Keywords' feature to select advertisers and is developing a pilot program with DoubleVerify and Integral Ad Science to assess brand suitability. The policy is to have independent companies verify compliance with ad safety standards while preventing access to users' private conversations.

Regarding the ad expansion, OpenAI stressed that "ChatGPT's responses remain independent, conversations are kept private, and users continue to control their own experience."

Implications for AX at Korean Companies

This test is limited to the US, and global expansion remains at the planning stage for OpenAI. For Korean companies, this isn't yet an advertising channel to act on immediately, but rather an early clue for what to prepare before it becomes available domestically.

If a marketing organization is considering ChatGPT advertising, the first step should be checking whether its systems are structured to pass conversion data to external ad channels, rather than focusing on impression counts. This is because OpenAI's stated measurement goal goes beyond clicks and conversions to incrementality. If a company plans to use the negative keyword feature, it should first compile a list of phrases and contexts in which its brand should not appear, in order to make effective use of that feature.

There's also a point worth considering from an AX (AI transformation for work) perspective. If the ad expansion centers on free and lower-tier plans, then which ChatGPT plan employees use for tasks like image generation becomes a starting point for gauging whether ads might appear on work screens. This week, take a moment to count how many people in your organization use ChatGPT for work and which plan each of them is on.

Source: OpenAI Introduces 'Visual Ads' to ChatGPT, Displayed Alongside Image Generation