OpenAI tests Sponsored Agents, letting ChatGPT users chat directly with advertisers

OpenAI is testing a new ad format called Sponsored Agents that lets ChatGPT users click into a labeled conversation with a business-sponsored AI chatbot directly inside the app — the clearest sign yet that ChatGPT's advertising business is moving from static ads toward interactive, agent-driven commerce. The company announced the test on September 16 with a small group of US advertisers.

How Sponsored Agents Work

After a user sees a relevant ad in ChatGPT, they can choose to open a clearly labeled conversation with that advertiser's own AI agent, separate from their original ChatGPT conversation and from ChatGPT's own independent answers. Inside that conversation, the user can describe their needs, ask follow-up questions, and get routed to the business's website when they're ready to act — closer to a live sales conversation than a static ad click. Home-services marketplace Angi is the first named launch partner, letting a user go from discussing a leaky faucet to booking a local contractor without leaving the chat.

Alongside Sponsored Agents, OpenAI is rolling out AI tools inside its Ads Manager for building campaigns automatically, plus direct integrations with HubSpot and Shopify — the Shopify integration expands internationally to every market where ChatGPT Ads is available starting September 23.

The Business Context

ChatGPT Ads is not a side experiment: OpenAI disclosed the business had already reached a $1 billion annualized revenue run rate as of August 31, 2026. Ads first appeared in the US in February 2026, expanded to the UK in June, and have since rolled out to Mexico, Brazil, Japan, and South Korea. Ads are shown only to users on the free tier and the low-cost Go plan — Plus, Pro, Business, and Enterprise subscribers remain ad-free, which keeps the advertising business dependent on converting free users rather than monetizing paying subscribers twice.

Why This Matters Beyond ChatGPT

Sponsored Agents represents a genuinely different advertising model than the search-ad or social-feed formats that built Google and Meta's revenue: instead of a ranked list of links or a scroll-stopping visual, the ad becomes a bounded conversation with a company's own AI, running inside the platform the user already trusts. That shifts the competitive question from "who shows the best ad" to "whose sponsored agent gives the most useful answer without feeling like a sales pitch" — a genuinely new skill for advertisers to build, and a new attack surface for the kind of manipulative or misleading agent behavior regulators have already flagged in unrelated AI-agent contexts.

OpenAI's decision to keep sponsored conversations clearly labeled and structurally separate from ChatGPT's own responses is a direct attempt to preempt the trust concern that killed early experiments with embedded advertising in other AI assistants: users need to be able to tell, at a glance, when they're talking to the product versus talking to an advertiser's chatbot wearing the product's interface.