OpenAI tests Sponsored Agents inside ChatGPT ads

OpenAI is testing a way for an advertiser to keep talking to you after you click its ad. In its announcement on September 16, 2026, the company introduced Sponsored Agents: click an ad in ChatGPT, and you can open a conversation with a chatbot the advertiser sponsors. The test is limited to select advertisers in the United States.

That turns an ad from a link into a conversation. For anyone whose product gets recommended inside ChatGPT, it also creates a paid version of the thing they were hoping to earn.

How a Sponsored Agent works

The flow starts with an ordinary ad. Unite.AI reports that a user who clicks one can opt into "a clearly labeled conversation with a business-sponsored agent," ask follow-up questions, and then click through to the business's site.

Two boundaries are the part to notice. OpenAI says these conversations are "distinct from ChatGPT's independent answers," and both outlets report they are kept separate from the user's main chat thread. Securities.io describes them as "distinct from regular ChatGPT interactions" with labeling to mark sponsored content.

The two reports differ on who is taking part. Unite.AI names Newegg, Best Buy, Lowe's and VistaPrint among early pilot advertisers. Securities.io says OpenAI disclosed no company names, describing the testers only as select advertisers in the United States. Treat the named list as one outlet's reporting rather than a confirmed roster.

What shipped alongside it

The agent pilot arrived with plumbing for the people who buy ads.

Item	What the reports say
Sponsored Agents	Test with select US advertisers, no timeline for wider release
HubSpot	First CRM partner; ads can be managed inside HubSpot from September 16, 2026
Shopify	Free ChatGPT Ads app that syncs inventory, builds campaigns and tracks performance
Text customization	Opt-in; adapts existing headlines and descriptions to the conversation, and translates copy
ChatGPT Ads reach	40 or more countries, per Securities.io; the agent test is US-only

The text feature is the one that changes what an ad is. Securities.io quotes OpenAI saying it "adapts an advertiser's existing headlines and descriptions" to fit a conversation's context. It also translates ad copy automatically. So the wording a user sees is generated per conversation, not written once and approved.

None of this arrives in a vacuum. This site reported that ChatGPT Ads hit a $1 billion annualized run rate in under 200 days as of August 31, a figure both of these reports repeat.

What this means for developers

There is nothing to build against yet. Neither report names an API, an SDK, an agent manifest or any documentation that would let a third party create a Sponsored Agent or query one. The integrations announced are for HubSpot and Shopify, not for your code. So this is a channel to understand, not a platform to adopt.

The question worth asking early is where your product sits when a competitor buys an agent. If a user asks ChatGPT for a recommendation in your category, there is now an organic answer and a paid conversation beside it. OpenAI's own wording says the two are distinct, and that distinction is the whole argument. Check it by hand in your own category rather than taking it on faith.

Per-conversation ad copy also breaks an assumption that most review processes rest on. If headlines are adapted to context and auto-translated, nobody signs off on the exact sentence a user reads. For regulated categories, or any claim your legal team has approved word by word, that is a real change. The opt-in switch is the control, and it is worth knowing who at your company owns it.

Watch the labeling, because it is the only thing separating a recommendation from an advertisement. Both reports quote OpenAI saying the conversations are clearly labeled. Look at what that label actually says in the product, and how visible it is on a phone. Then check whether a user three messages deep still knows which of the two they are talking to.