OpenAI started testing ads in ChatGPT on February 9, 2026. They show up below responses for Free and Go plan users in the US. Sponsored labels. Separate ad systems from the chat model. Paid tiers (Plus, Pro, Business, Enterprise, Edu) stay ad-free. OpenAI says advertisers can't shape ChatGPT's responses. What they can do: use your chat content to pick which ads you see. The system matches ads to your current conversation topic. Opt into personalization and it pulls from past chats and memory too. Sarah Bird joined as Chief Marketing Officer to run this, recruited from Meta and Google. OpenAI has been hiring ad talent from Big Tech aggressively. That tells you something. Five guiding principles for the program: mission alignment, answer independence, conversation privacy, user control, long-term value. No ads near sensitive topics like health or politics. Advertisers get aggregated performance data. Never your actual conversations. Agent builders take note: this is your roadmap or your cautionary tale. OpenAI's bet is that walling off ad infrastructure from model outputs keeps trust intact. Maybe it works at first. When ad revenue becomes real money, that wall gets tested. It happened at Facebook, Google, Twitter. The wall always gets tested.