Summary: OpenAI’s ads portal reached wider attention on July 21 as a self-serve Ads Manager beta for ChatGPT advertising. Earlier tests began on February 9 with logged-in adults on Free and Go tiers in the United States, while Plus and Pro subscribers see no ads. The page also commits to clear labeling, separating ads from answers, and user control over ad data use.
Advertisers can now handle this themselves. On July 21 OpenAI’s ad portal reached the Hacker News front page, gathering over a thousand points and more than eight hundred comments. Located at ads.openai.com, that portal moves the ChatGPT advertising business into its next phase: a self-serve Ads Manager, beta label and all, has become the front door to conversational traffic. Registration, account setup, campaign launch — advertisers do all of it unaided. Against February’s closed test, the self-serve platform amounts to a declaration that OpenAI’s ad operation now stands open to budgets of every size.
How a February Pilot Became Self-Serve
Advertising has sat inside ChatGPT for months, so the concept alone is hardly news. OpenAI announced the program in January; sponsored content went into testing on February 9 with logged-in adult users on the Free and Go tiers in the United States, while paid subscribers on Plus and Pro see none of it. Criteo signed on in early March as the first ad tech partner and brought roughly 17,000 advertisers into the pilot, and by the end of March the test had expanded into Canada, Australia, and New Zealand.
The shift this time sits on the supply side. Access once came through a partner referral or a sales contact; now an advertiser registers at the portal and opens an Ads Manager account unaided, with a starter offer of $500 in ad credit against a $500 spend on the table. Invitation-only giving way to self-serve says plainly that OpenAI wants the long tail of budgets and means to run advertising as a revenue pillar beside subscriptions.
Inside the Ads Manager Workflow
OpenAI’s onboarding material condenses the work into three steps: create a campaign with its budget and goal; add ad details, by bulk upload or within the tool; then launch, refining as measurement and performance come in. The Help Center now carries an Ads Manager Beta overview and quickstart guides too. For anyone who has bought Google or Meta inventory, the interface holds few surprises. Where it differs is the target — someone mid-conversation, rather than a keyword or an audience segment.
Deeper-pocketed advertisers are courted separately: larger accounts with more complex needs get dedicated sales support, while the self-serve lane serves the rest. The customer stories skew direct-response as well — Newegg’s testimonial comes from a performance marketing manager rather than a brand team, which hints that the first budgets gear toward conversions instead of awareness.
The Sales Pitch: Reaching People Mid-Conversation
The portal does not dress up its pitch. A visit to ChatGPT is not merely a hunt for information, the page argues: people explore options, compare alternatives, weigh tradeoffs, and decide — and that is where advertisers get to stand at “moments of real intent.” Conversations, it adds, carry richer context signals too, and those are what the company credits for advertising that lands as more relevant, more personalized, and more useful.
Four names appear on the site already: Newegg and Best Buy, plus Lowe’s and VistaPrint. Alec Shao, who runs performance marketing at Newegg, describes the aim this way — when somebody is researching what to buy, Newegg wants to be present in that conversation, capturing the consideration window.
Trust Language, and Regulatory Heat
OpenAI sets out three principles on the page itself: each ad is identified clearly; advertising stays separate from ChatGPT’s answers; and users choose whether their data is used in ads. Trust language at the front door rests on a regulatory backdrop — in late January, Senator Markey wrote to OpenAI and six other AI companies, asking what chatbot ad plans imply for privacy, consumer protection, and teen safety.
On the commercial side, February’s two-track arrangement is unchanged: ads show up on Free and Go, and paying users see none. OpenAI treats a subscription as an experience promise and advertising as the way free traffic earns. Loosen that line, and the paywall’s sales pitch loses force.
An Advertiser’s or Developer’s Read
Two practical judgments follow. The first concerns traffic quality, where early evidence is encouraging but far from settled: Criteo reported in March that LLM traffic converts at roughly 1.5 times the rate of other channels. Now that self-serve is live, a small budget spent probing conversational placements and creative tone is a reasonable way to begin. The second is that the rules will move again — the beta tag, how quickly each country comes online, and where the free tier ends and paid begins are all liable to change over the next few quarters. Booking ChatGPT placements under experimental budget instead of treating them as a core channel stays the steadier posture.
Competition only raises the stakes. Perplexity has sold search ads since 2025, Meta keeps weaving advertising into its AI surfaces, and every assistant maker is after a second revenue stream beyond subscriptions. Whether conversation-native advertising scales is a question ChatGPT now answers on a bigger stage than anyone has tried — and its handling of labeling, separation, and user control will likely become the reference points that regulators and competitors borrow.