OpenAI tests labeled ads inside ChatGPT for free and Go users

OpenAI has begun limited tests of advertisements inside the ChatGPT app that will appear beneath the AI’s answers for nonpaying users and subscribers to its lower-priced Go plan. The company says the placements will be clearly labeled and separated from the assistant’s responses, and that the trials will be limited in scope and rolled out in the coming weeks in the United States.

Screenshots shared by the company show compact banner-style boxes with a small image and brief copy blocked off from the body of the AI’s reply. OpenAI says ads will not alter how ChatGPT generates answers and that conversational content will not be shared with marketers. The company also says it will not show ads to users under 18 and will avoid placing promotions alongside sensitive topics such as mental health, medical queries and political content.

The pilot targets logged-in users on the free tier and the new ChatGPT Go plan, which is being offered in the U.S. at about $8 per month. Higher-priced subscriptions will remain ad-free; customers on Plus, Pro, Business and Enterprise plans will not see the trial ads. OpenAI is pitching the move as a way to broaden access to the service and diversify revenue streams as it faces the rising costs of running data centers and training advanced models.

OpenAI is also building controls meant to increase transparency and user choice. Reported features include labels that explain why a particular ad was shown, options to dismiss ads and submit feedback, and settings to turn off ad personalization and clear ad-related data. Company materials indicate ads would be contextually relevant to a conversation, asking for travel recommendations, for example, could yield a sponsored hotel or holiday offer beneath the AI’s suggestions.

The change follows weeks of user frustration after ad-like elements briefly appeared in conversations, a development that revived earlier warnings from the company’s leadership. Sam Altman said in 2024 that advertising in ChatGPT was a "last resort", and the new tests mark a partial reversal of that stance. Industry estimates suggest roughly 95 percent of ChatGPT users rely on the free tier, underscoring the scale of the audience that could encounter ads if the tests expand.

Analysts and privacy advocates warn the introduction of ads could erode user trust and prompt some people to migrate to alternative services. OpenAI counters that preserving ad-free experiences for paying customers creates a clear incentive to subscribe while enabling the company to monetize broader access without harvesting conversational data for advertisers.

The experiments also raise questions for regulators and advertisers about how AI platforms blend commercial content with conversational interfaces. If the tests proceed to a wider rollout, the effectiveness of contextual ad placements in a generative AI setting will be closely watched by marketers betting that smarter recommendations can boost performance in search-like and social channels. For users, the test will be an early measure of whether labeled ads can coexist with the immediacy and perceived neutrality of AI assistants.