September 25, 20269 min read

ChatGPT Sponsored Agents, explained: how ads inside AI agents work and what they mean for users

What OpenAI's Sponsored Agents are, who sees ChatGPT ads, what data targeting uses, how to turn ads off, and the open questions about ads inside agents.

ByLokesh Kumar· Founder & Builder, Vyra
Share on XShare on LinkedIn

The short answer

Sponsored Agents are a new ChatGPT ad format: after clicking an ad, you can open a labeled chat with an AI agent built for that business. OpenAI announced them on 16 September 2026 as a test with select US advertisers. They appear only where ChatGPT shows ads, which means the Free and Go plans.

What is a ChatGPT Sponsored Agent?

It is an advertiser's chatbot that runs inside ChatGPT. OpenAI's help article describes it as "an advertising format that lets people chat with an AI representative of a business in ChatGPT."

The flow, according to OpenAI's announcement, is:

  1. You are chatting with ChatGPT and a relevant ad appears.
  2. You choose to open a conversation with the business-sponsored agent.
  3. You "explain what matters to them, ask follow-up questions, and follow a link to the business's website when they're ready to take the next step."

OpenAI says the Sponsored Agent conversation "is distinct from ChatGPT's independent answers and separate from the original conversation that the user started in ChatGPT." In other words, it is a new thread with a different party, not ChatGPT changing its own answer.

TechRepublic reports that OpenAI said it is building these with "a limited set of initial testing partners," giving advertisers the ability to "shape and finetune the agent" based on their existing websites and services, and that "Sponsored Agents do not support purchases." That last point is not in OpenAI's own post, and at least one other outlet describes purchases as a possibility, so treat checkout as unsettled. OpenAI's own wording only covers sending you to the business's website.

The help article says access is "currently available only to selected advertisers in a limited alpha test." This is a pilot, not a general launch.

The same announcement also added an Ads Manager plugin for advertisers and integrations with HubSpot and Shopify. Those are advertiser-side tools and do not change what users see directly. If the distinction between a chatbot and an agent that acts is fuzzy, our explainer on AI agents vs chatbots covers it.

Where do ChatGPT ads appear, and how are they labelled?

Standard ChatGPT ads came first. OpenAI's January post on its approach to advertising said it would test ads "at the bottom of answers in ChatGPT when there's a relevant sponsored product or service based on your current conversation," and that ads would be "clearly labeled and separated" from the answer.

Sponsored Agents build on that ad unit. The ad is the entry point, and the agent chat opens only if you choose it. OpenAI calls the resulting conversation "clearly labeled," but neither the announcement nor the help article specifies the exact label text or visual treatment. We have not been able to verify the UI firsthand, since the test is limited to select US advertisers.

Which plans and countries see ChatGPT ads?

QuestionWhat OpenAI saysSource
Plans with adsFree and GoAds in ChatGPT help
Plans without adsPlus, Pro, Business, Enterprise, EduAds in ChatGPT help
MinorsNo ads for accounts identified as under 18Ads in ChatGPT help
First testUS, logged-in adults, from 9 February 2026Testing ads in ChatGPT
Current reach"over 60 countries" as of 23 September 2026OpenAI SE Asia post
Sponsored AgentsUS only, select advertisersOpenAI announcement

The geographic expansion has been fast. OpenAI announced 31 European countries in August, then Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan on 23 September. Sponsored Agents, however, are still described only as a US test.

What data does OpenAI use to target ChatGPT ads?

This is where the details matter. According to the Ads in ChatGPT help article, when ad personalization is on, ads can draw on:

OpenAI's February testing post describes the same thing in plainer terms: "matching ads submitted by advertisers with the topic of your conversation, your past chats, and past interactions with ads."

What OpenAI says advertisers do not get, per the help article: "your chats, chat history, memories, name, email, precise location, IP address, or sensitive information." Advertisers "receive aggregated reporting such as views and clicks." OpenAI also says it "never sell[s] your data to advertisers" (January post) and that ads are not eligible near "personal health, mental health, or politics."

So the accurate framing is: your chats are used by OpenAI to pick ads, but OpenAI says they are not handed to advertisers. That is roughly the model search and social ad platforms use. The difference is that a chat history with an assistant tends to hold more context than a search log. We covered why that matters in AI assistant privacy and local-first design.

What happens to what you tell a Sponsored Agent?

This is the gap in the current documentation. The privacy commitments above were written for ads shown next to ChatGPT's answers. A Sponsored Agent is different: you are typing directly into a conversation whose purpose is to represent a business.

Neither OpenAI's announcement nor the Sponsored Agents help article states whether the business can read transcripts of those conversations, what the business receives beyond aggregate metrics, whether context from your original chat is passed to the agent, or how long the conversation is retained. TechRepublic also notes the advertiser-access question is not addressed.

That may be answered as the test expands. Until it is, a reasonable default is to treat a Sponsored Agent like a sales chat widget on a company website: useful for product questions, not the place for details you would not tell a salesperson.

Is it a conflict of interest when an agent is also an ad surface?

OpenAI's stated line is that "ads do not influence the answers ChatGPT gives you" (January post). Sponsored Agents are structured to keep that line intact: the paid agent lives in a separate, labeled conversation rather than inside ChatGPT's answer. That is a defensible design, and it is fair to credit it.

The harder question is what happens as assistants move from answering to acting. An agent that compares prices, fills carts, or books things on your behalf is making decisions, and every point where a paid party can enter that loop is a place where your interest and the platform's revenue can diverge. Two recent stories show the pressure from both sides:

Neither story is evidence that Sponsored Agents misbehave. They are context for why disclosure and separation matter more for agents than for a banner ad. For a broader look at limiting what an agent can do on your behalf, see human-in-the-loop AI agents and AI agent security and blast radius.

Common questions about ChatGPT Sponsored Agents

Can I turn off ads in ChatGPT?

Yes, in two ways, according to OpenAI. Plus, Pro, Business, Enterprise and Edu plans have no ads. On the Free plan, the help article says you can choose an ad-free option in exchange for lower usage limits and reduced feature access, such as fewer messages and no image generation.

Are ChatGPT Sponsored Agents safe?

OpenAI says they are clearly labeled, separate from ChatGPT's own answers, and only opened when you choose to click. What it has not yet published is how data you type into a Sponsored Agent is shared with the business or retained. Treat them as you would a company's sales chatbot and avoid sharing sensitive personal details.

Does ChatGPT use my chats for ads?

With ad personalization on, yes: OpenAI says ads may use your current chat, past chats, memory, and ad interactions to choose what to show. OpenAI says advertisers never receive your chats, memories, name, email, precise location or IP address, and get only aggregated reporting. You can turn personalization off and delete ad data, which OpenAI says is retained for up to 30 days.

Do ads change ChatGPT's answers?

OpenAI says no: "Ads do not influence the answers ChatGPT gives you." Ads appear below answers, and Sponsored Agents run in a separate conversation. There is no independent audit of this claim that we could find.

Can I get Sponsored Agents outside the US?

Not yet, based on OpenAI's announcement, which describes testing "with select advertisers in the United States." Standard ChatGPT ads are live in over 60 countries.

Where Vyra by Vyraagi stands

Vyra by Vyraagi is a desktop AI agent in closed alpha. Its episodic memory and structured world model are held on-device, and its 5-tier model router can fall through to a local Ollama tier when offline, which keeps the question "who can read my history?" closer to home. Its metacognition layer is designed to detect irreversible actions before they execute, the kind of check we think matters more as agents start acting on your behalf.

Sources

Vyra by Vyraagi is building a desktop agent that keeps its memory on your machine. If that's what you want, join the waitlist.

Vyra is in closed alpha now, with a Founders Beta ahead of public launch.

Join the WaitlistSee Plans

Related reading

How much memory do you need to run local LLMs in 2026? (And why Mac minis keep selling out)

Weights, KV cache and OS overhead, worked out from bits per weight. A size table for 3B to 70B models and why bandwidth sets tokens per second.

Does the New Siri Send Your Data to Google? How iOS 27's Gemini-Based Siri Handles Privacy

iOS 27's Siri is built on Google's Gemini and partly runs on Google Cloud. What Apple says reaches Google, what doesn't, and the settings that control Siri AI.

Small Language Models vs LLMs: Why the Next AI Wave Is Getting Smaller

Small language models are 10–30× cheaper to run than LLMs and could handle 40–70% of an agent's work. When small beats big, when it doesn't, and why.