On February 9, 2026, OpenAI began testing “sponsored” ads in ChatGPT. The scope is precise: logged-in adult US users on the Free and Go tiers. The test had been announced on January 16; the live testing started February 9.
Paying users are outside the experiment — Plus, Pro, Business, Enterprise, and Education plans see no ads.
Who Sees Ads and Who Does Not
The line OpenAI drew is itself the business model:
- Sees ads: logged-in adult US users on Free and Go
- Sees no ads: Plus, Pro, Business, Enterprise, Education
Translated into commercial language: the subscription fee buys silence, and the free tier’s attention gets monetized through advertising. The dual track that has run the consumer internet for two decades now lands on an AI assistant.
One wrinkle is the Go tier. Go sits above Free but below Plus — a paying tier that still sees ads breaks the simple equation of subscription-equals-silence. OpenAI is effectively running three monetization modes at once: ads only, subscription only, and a hybrid in between. That hybrid is where the pricing psychology gets interesting, and where complaints will land first if the ad load feels heavy.
The Free Tier’s New Contract
ChatGPT’s free tier has always been OpenAI’s largest acquisition asset — enormous usage, continuous inference costs, and users who have not converted to subscriptions. Advertising is a third monetization path, and it aims directly at the biggest layer of the user base.
The hard part is the product. Search engines spent twenty years building placement conventions; users know which block is an ad before they read a word. A conversational interface has no such convention. What form sponsored content takes, how often it appears, and how clearly it is labeled will decide how fast user trust erodes. That is likely why OpenAI chose a test rather than a full rollout — the open variable is acceptance, not technology.
There is also a trust asymmetry at play. ChatGPT built its habit on being a direct answer machine; a results page trained users to scan past ads, but nothing has trained them to discount parts of a conversation. If sponsored content reads as part of the answer rather than alongside it, the cost of this test will be measured in trust, not click-through.
What to Watch
The test has just started; three things are worth tracking:
- The actual form and frequency of sponsored content in conversations, and how users respond
- Whether the test expands beyond Free and Go, or beyond the US
- Whether advertising revenue starts playing a louder role in OpenAI’s commercial narrative
For product teams the takeaway is direct: if your product benchmarks itself against ChatGPT, user expectations now carry an advertising variable. For marketing teams, a new placement surface is taking shape. The comparison baseline is well known: search advertising works because intent is explicit. Whether conversational ads can reach similar economics without degrading the core product is genuinely unsolved — this test is the first large-scale data point.
Sources
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.