Do ads belong in ChatGPT? OpenAI’s latest move suggests that the answer is “maybe, but only very carefully.” On January 16, 2026, OpenAI said it would begin testing ads in the United States to help keep ChatGPT accessible on free and low-cost plans. Then, on February 9, 2026, the company started that test for logged-in adult users on the Free and Go tiers. Plus, Pro, Business, Enterprise, and Education users do not see ads. (openai.com)
The design is important. OpenAI says ads appear below the end of a response, are clearly labeled as sponsored, and do not change ChatGPT’s answers. Advertisers do not get access to users’ chats, memories, or personal details; they receive only aggregated performance data such as views or clicks. Users can dismiss ads, turn off ad personalization, clear ad data, and even choose an ads-free version of the Free plan with lower message limits and fewer features. During this early test, ads also do not appear in Temporary Chats, for logged-out users, or near sensitive topics such as health, mental health, and politics. (openai.com)
What makes this more than a normal ad test is OpenAI’s bigger idea. On its advertiser page, OpenAI says people use ChatGPT not only to ask questions, but also to compare options and decide what to do next. The company also says future ads may become interactive, allowing users to ask follow-up questions before making a purchase decision. This suggests the next stage of marketing may be “conversational”: less like a banner on a webpage, and more like a guided discussion inside the moment of decision. (openai.com)
Still, the risks are serious. Academic research has found that people can struggle to notice ads when they are blended into chatbot replies, and another study warns that conversational systems may create a “fake friend” problem, where users trust the assistant too much. That is why OpenAI’s promise of clear labels, strong controls, and answer independence matters so much. In short, ads may fit ChatGPT only if they stay visibly separate from help itself. If that line becomes blurry, user trust could disappear very quickly. (arxiv.org)










