OpenAI Tests Sponsored Agents as ChatGPT Ads System Matures
OpenAI is testing Sponsored Agents inside ChatGPT, a format that lets a user start a conversation with a business-sponsored agent after clicking an ad.
The test is limited to select advertisers in the United States. Conversations with Sponsored Agents are clearly labeled. A conversation with a Sponsored Agent is distinct from ChatGPT's independent answers and separate from the user's original ChatGPT conversation, according to OpenAI.
That separation matters for implementation. The ad click does not continue the existing thread. It opens a different, labeled conversational surface controlled as a sponsored experience. For developers and media buyers accustomed to landing pages and click IDs, the unit here is a dialogue, not a destination URL.
The Sponsored Agents test sits inside a larger update to how campaigns are built and managed. Advertisers can use natural-language prompts to create, update, and analyze campaigns directly in ChatGPT with the Ads Manager plugin. The workflow stays in chat. A marketer describes an objective, asks for changes, or requests performance analysis without leaving the ChatGPT interface.
OpenAI has also added AI assistance in Ads Manager that suggests ad copy and imagery based on landing page and campaign objective. A separate opt-in text customization feature adapts headlines and descriptions to conversation context and translates ad copy to a user's preferred language. The first capability is generative assistance at build time. The second is dynamic assembly at serve time, tuned to the surrounding conversation.
Distribution is moving through existing SaaS surfaces. HubSpot is OpenAI's first CRM partner for ChatGPT Ads integration. Businesses can connect a ChatGPT Ads account in HubSpot to create ads, track performance, and follow up on leads directly in HubSpot. Shopify is OpenAI's first ecommerce partner for ChatGPT Ads integration. US-based Shopify merchants can use the new ChatGPT Ads app in the Shopify App Store to create and manage ChatGPT ad campaigns.
OpenAI lists the update as 'Reimagining advertising with AI' on its OpenAI News page, where it is dated Sep 16, 2026 and categorized as Product. The company also hosts a page titled 'ChatGPT Ads, a new channel for marketers' on its webinar subdomain.
Scale and expansion
The Sponsored Agents test follows rapid scaling of the underlying ads business. ChatGPT Ads has reached a $1 billion annualized revenue run rate, OpenAI said on Aug. 31. ChatGPT Ads will expand to 31 European countries, including Germany, France, Spain, and Italy, according to an Aug. 18 update.
Earlier details trace the ramp. OpenAI planned to test ads at the bottom of answers in ChatGPT when there is a relevant sponsored product or service based on the current conversation. On May 7, 2026, the company said it planned to expand the ChatGPT ads pilot in the coming weeks to the United Kingdom, Mexico, Brazil, and Japan. It later planned to show ads to all users of the free and Go versions of ChatGPT in the United States in the coming weeks.
The US pilot expanded to over 600 advertisers and exceeded $100 million in annualized revenue within its first six weeks. Nearly 80% of small- and medium-sized businesses signaled interest in ChatGPT ads in that period.
What changes for practitioners
The broader context here is a shift from retrieval-style sponsored links to conversational inventory. Search ads matched keywords to bids. Social ads matched identity graphs to feeds. ChatGPT ads match conversation context to creative variants, and Sponsored Agents extend that logic one step further by making the post-click a conversation.
In this author's view, the operational questions will be familiar to anyone who lived through the PC, web, mobile, and cloud transitions. Measurement comes first. Prompt-built campaigns, context-adapted headlines, and translated variants will complicate creative versioning and attribution unless HubSpot-style closed-loop tracking and Shopify-side transaction data carry their weight. Brand safety comes second. Clear labeling and thread separation help, but sponsored dialogue raises new edge cases around disclosure persistence, refusal behavior, and how far a sponsored agent can go in troubleshooting, recommending, or transacting before it must hand off.
Worth flagging for enterprise teams is the direction of the tooling. Natural-language campaign management lowers the barrier to launch. It also moves tacit knowledge about audiences, budgets, and creative testing into prompts and plugin state. Teams that document those prompts, pin successful variants, and audit AI-suggested copy and imagery with the same rigor they apply to code review will adapt faster than teams that treat the chat box as a shortcut.
I have watched my own children move from typing queries to expecting answers to expecting action, often without noticing the shift. That is the adoption pattern to watch here. If sponsored conversations prove useful for narrow tasks like sizing, compatibility checks, or rebooking, users may accept them in the same way they accepted sponsored results once those results solved a problem. If they feel evasive or overeager, labeling alone will not save them.
The technology arc still points toward more capable assistance, and conversational commerce is a plausible beneficiary. Sponsored Agents are early, limited, and clearly marked. The infrastructure around them, from Ads Manager prompts to CRM and storefront integrations, suggests OpenAI is building for campaigns that are authored, served, measured, and fulfilled without leaving a dialogue.
