Summary: OpenAI began live-testing sponsored ads in ChatGPT on February 9, 2026, limited to signed-in adults in the US on Free or Go plans. Plus, Pro, Business, Enterprise, and Education accounts were excluded, although Go remains the paid exception that still receives ads. The pilot’s unresolved question is whether conversational placements can earn acceptance without eroding trust in direct answers.
February 9, 2026 was the day OpenAI put “sponsored” advertising into live testing inside ChatGPT. Inclusion is narrow: adults in the US, signed in, on the Free or Go plan. The experiment was flagged on January 16; the run itself began February 9.
Subscribers are excluded: Plus, Pro, Business, Enterprise, and Education accounts won’t see a single ad.
Where the Ad Line Sits
OpenAI’s boundary pretty much declares the business model on its own:
- Ad viewers: logged-in adult US users on Free and Go
- Ad-free: Plus, Pro, Business, Enterprise, Education
In commercial terms, subscribers buy quiet; the free tier’s attention becomes inventory that advertising monetizes. The two-track arrangement consumer internet businesses used for two decades now reaches an AI assistant.
Go is the odd one out: above Free yet below Plus, it is a paid tier that still receives ads, puncturing the tidy assumption that subscribing equals silence. So OpenAI juggles three monetization models at once — pure advertising, pure subscription, and a blend between them. That blend is where pricing psychology turns interesting and a heavy ad load would draw the first complaints.
The Free Tier’s Side of the Deal
For OpenAI, ChatGPT’s free tier has long been the biggest acquisition engine — colossal usage, ever-accruing inference costs, and many users who never converted to paid. Advertising opens a third revenue lane aimed at the widest slice of the user base.
The difficult piece is product design. Twenty years gave search engines settled placement conventions; readers spot an ad block before reading a word, while conversational interfaces carry no such shared grammar. The shape sponsored content takes, its frequency, and its labeling legibility set the pace at which user trust wears down. Presumably that is why OpenAI went with a test instead of an immediate launch — the unresolved variable is acceptance, not technology.
A trust asymmetry sits under all of this. The ChatGPT habit formed around a direct answer machine; a search results page taught users to skim past ads, yet nothing conditioned them to doubt a passage inside a conversation. When sponsored content reads like part of the answer instead of something beside it, what this experiment spends is trust, not click-through.
Three Things Worth Tracking
The experiment is barely underway; these questions are worth following:
- What sponsored content actually looks like in conversations, how often it appears, how users respond
- Whether the rollout stretches past Free and Go, or past US borders
- Whether ad revenue gains a louder voice in the commercial story OpenAI tells
Product teams get a blunt lesson: benchmark against ChatGPT and user expectations now include an ad variable. Marketing teams watch a fresh placement surface assemble itself. The reference case is familiar: search advertising succeeds on explicit intent. Whether conversational ads can hit comparable economics without corroding the core product is genuinely unsolved; this test provides the first large-scale data point.