OpenAI announced ads in ChatGPT on January 16, 2026. Testing went live on February 9, 2026. If you haven’t looked up from your Google Ads dashboard since then, you’re already behind — not because ChatGPT advertising is proven performance territory yet, but because the underlying question has been answered: this channel is real.
The question everyone in paid media should be asking right now isn’t “will OpenAI ever run ads?” It’s “what exactly are these ads, how do they work, and what do I need to know before this beta opens up?” That’s what this article answers — with verified facts only, no speculation dressed up as insight.
- OpenAI began testing ads in ChatGPT on February 9, 2026 — only for US users on the Free and ChatGPT Go ($8/mo) tiers.
- The only confirmed ad format is a single chat_card: a titled, image-backed sponsored card placed below ChatGPT’s answer, clearly labeled “Sponsored.”
- OpenAI explicitly states that ads do not influence the assistant’s response — the answer comes first, the ad appears after.
- Paid tiers (Plus, Pro, Business, Enterprise, Education) and users under 18 do not see ads.
- This is an early beta on a fast-evolving channel. The fundamentals are documented — the performance data is still thin, and any agency claiming otherwise is guessing.
The Short Version: What Are ChatGPT Ads, Exactly?
ChatGPT ads — also called OpenAI ads or ads in ChatGPT — are sponsored placements that appear within the ChatGPT interface after the AI delivers its answer to a user’s query. They are not injected into the response itself. They sit below it, in a clearly labeled card format, separated from the organic content.
Think of it less like a Google search ad (which appears above results) and more like a contextually triggered card that shows up once the conversation has already happened. The user asked something. ChatGPT answered it. Then, if the query context matches an advertiser’s targeting, a sponsored card appears beneath.
This distinction matters. OpenAI has been explicit that the assistant’s answer is not shaped by advertising. Whatever ChatGPT tells you is independent of who’s paying. The ad is downstream of the response, not embedded in it. That architecture is likely a deliberate trust decision — and a smart one, given how quickly users would revolt if they suspected the AI was steering recommendations toward paying sponsors.
The Only Confirmed Format: The chat_card
Right now, there is one documented ad format in ChatGPT: the sponsored chat card, identified in OpenAI’s ad spec as chat_card.
Here’s what a chat_card contains:
- Title: 3–50 characters
- Body copy: approximately 100 characters
- Image: a visual asset accompanying the card
- Favicon: the advertiser’s site icon for brand recognition
- Link: a destination URL
- Sponsored label: visible disclosure that this is a paid placement
That’s the full spec as documented. It’s tight — 50 characters for a headline makes Google’s RSA format look generous. Writing effective copy for this format is a real craft challenge, and it rewards specificity over cleverness.
One thing worth emphasizing: there are no confirmed video formats, no confirmed carousel formats, no confirmed in-response placements as of this writing. If you’re reading somewhere that ChatGPT is running video ads or multi-card sequences, that’s speculation. The only format OpenAI has documented and deployed in this beta is the chat_card. This channel is evolving fast, and new formats will almost certainly emerge — but we’re only writing about what’s real today.
Who Sees ChatGPT Ads (And Who Definitely Doesn’t)
This is one of the most important things to understand about the current state of ChatGPT advertising, because it shapes the realistic audience size and targeting picture significantly.
Who sees ads in ChatGPT right now:
- US users on the Free tier
- US users on ChatGPT Go ($8/month)
Who does not see ads:
- ChatGPT Plus subscribers
- ChatGPT Pro subscribers
- ChatGPT Business accounts
- ChatGPT Enterprise accounts
- ChatGPT Education accounts
- Users under 18, regardless of tier
- Non-US users (in this beta phase)
That’s a meaningful constraint. The paying subscriber base — the people most invested in ChatGPT as a daily tool — isn’t seeing your ads. You’re reaching free users and Go subscribers. That’s still a massive audience given ChatGPT’s scale, but it’s a different audience profile than you might assume when you hear “advertise on ChatGPT.”
Whether OpenAI expands ad exposure to more tiers over time is genuinely unknown. The economics suggest they will. The trust dynamics suggest they’ll move carefully. Watch this space.
How ChatGPT Ads Differ From What You Know in Google and Meta
If you’ve managed Google Ads at any serious budget, your mental model of paid search is built on keyword auctions, Quality Scores, and intent signals derived from the exact words someone typed. Google Ads and Facebook Ads operate on fundamentally different intent models — search vs. interest — and ChatGPT advertising introduces a third paradigm entirely.
In a ChatGPT conversation, users don’t type keywords. They type questions, describe problems, explain contexts. “I’m planning to redo my kitchen in a contemporary style on a $30K budget, what should I prioritize?” That’s not a keyword. It’s a conversation signal — richer in context than any search query, but structured differently than anything Google’s auction has ever ingested.
What OpenAI does with that conversational context to trigger ad placements is not fully public. The targeting mechanics behind ChatGPT advertising are still being documented as the beta evolves. What we know is that the ad appears when the conversation context is deemed relevant to the advertiser’s offering. How exactly that matching works — whether it’s topic-based, intent-based, keyword-derived from the conversation, or some combination — is an area where the industry is still learning. We cover the targeting mechanics in more depth in our dedicated piece on how ChatGPT ads targeting and audience selection works.
The placement position is also different from Google. In paid search, your ad competes to appear before organic results. In ChatGPT, the AI’s answer always comes first. The ad is below. This isn’t a bug — it’s the architecture OpenAI chose to protect user trust. It also means the user has already received value before they see your card, which could mean higher intent at the point of exposure, or it could mean lower urgency to click. We don’t have enough data yet to say definitively.
Is This the Same as ChatGPT Ads Being “Self-Serve”?
Yes — ChatGPT Ads launched as a self-serve advertising platform in 2026. This puts it alongside Google Ads and Meta Ads as a channel advertisers can access and manage directly, without going through a managed deal or insertion order.
That’s significant. It means any advertiser, at almost any budget, can theoretically run OpenAI ads through a self-serve interface. It also means the channel will get crowded faster than managed-only platforms, and the advertisers who build expertise early will have a real structural advantage before CPCs normalize.
For context on how that early-mover dynamic has played out elsewhere: when Google’s self-serve auction opened up, the advertisers who understood Quality Score and account structure before the masses arrived got disproportionate returns. The same pattern tends to repeat on every new paid channel. The question is whether you want to be learning before the crowd shows up, or after.
If you want a direct comparison of how the two dominant channels stack up right now, our article on ChatGPT Ads vs Google Ads walks through the honest tradeoffs without the hype.
What OpenAI Has Said About Ad Transparency and Response Independence
OpenAI has made one commitment publicly that every advertiser — and every user — should understand: ads do not influence ChatGPT’s answers.
The assistant responds to queries based on its training and reasoning. A sponsored card placed below the answer does not change what the AI said. An advertiser cannot pay to have ChatGPT recommend their product, steer users toward their service, or omit competitors from its response. The answer is organic. The card is paid. They are separate.
This is a structural decision that OpenAI has been deliberate about, and it makes sense. The moment users suspect the AI is for sale in its recommendations, the trust that makes ChatGPT valuable collapses. So the sponsored chat card is downstream of the answer — not embedded in it — and the “Sponsored” label is visible.
That said, this is an early beta, and OpenAI’s policies and implementations will evolve. Advertisers should watch any updates to these guidelines carefully, because shifts in how the platform handles the relationship between paid content and AI responses would have significant implications for how the channel functions and how much users trust it.
Should Your Business Pay Attention to This Right Now?
Honest answer: it depends on your risk tolerance, your available budget for channel experimentation, and what you’re selling.
If you’re a direct-to-consumer brand with a strong visual product and a budget that can absorb some testing cost, getting familiar with the chat_card format now — before CPCs reflect a mature, competitive auction — is a reasonable bet. The audience is enormous even in the free tier, and the conversational context around commercial queries is genuinely rich.
If you’re a B2B company selling to enterprise buyers, the audience restriction (free and Go users only, US only) limits the channel’s near-term relevance. Your buyer probably isn’t a free-tier ChatGPT user. That might change as the platform evolves, but right now, your Google Ads and LinkedIn investment is almost certainly better allocated. Our breakdown of ChatGPT ads for B2B companies goes deeper on why the intent signals are promising but the audience composition matters.
If you’re allocating a fixed budget and trying to decide between established channels and this new one, don’t abandon what’s working. ChatGPT advertising is a channel to add and test, not a reason to pull budget from campaigns that are already generating ROI. The advertisers who got burned on the last “Google Killer” all made the same mistake: they rotated budget toward the new thing before the new thing had proven itself.
Frequently Asked Questions About ChatGPT Ads
When did ChatGPT ads start?
OpenAI announced the ads program on January 16, 2026, and began testing live ads on February 9, 2026. The rollout started with US users on the Free and ChatGPT Go tiers.
What does a ChatGPT ad look like?
The only confirmed format is the chat_card: a sponsored card with a short title (3–50 characters), approximately 100 characters of body copy, an image, a favicon, and a destination link. It appears below ChatGPT’s answer and is labeled “Sponsored.” There are no confirmed video or carousel formats in the current beta.
Do ChatGPT ads affect what the AI says?
No. OpenAI has explicitly stated that ads do not influence the assistant’s response. The AI answers first, independently. The sponsored card appears after the answer. Advertisers cannot pay to shape, steer, or influence what ChatGPT tells users.
Who sees ads in ChatGPT?
Currently, only US users on the Free tier and the ChatGPT Go ($8/month) tier see ads. Plus, Pro, Business, Enterprise, and Education subscribers do not see ads. Users under 18 are excluded regardless of tier.
Can any advertiser run ChatGPT ads?
Yes — ChatGPT Ads launched as a self-serve platform, meaning advertisers can access it directly without going through a managed deal. This puts it in the same accessibility tier as Google Ads and Meta Ads.
How is ChatGPT advertising different from Google Ads?
Google Ads are triggered by specific keywords typed into a search box, and they appear before organic results. ChatGPT ads are triggered by conversational context and appear after the AI’s answer. The intent signals are richer but structured differently, the audience is defined by platform tier rather than search behavior, and the format constraints are much tighter. For a full breakdown, see our comparison of how ChatGPT ads differ from traditional search ads.
Is this channel worth testing in 2026?
For consumer-facing brands with a budget for channel experimentation: yes, it’s worth understanding and testing in controlled fashion. For B2B advertisers targeting senior buyers, the current audience constraints limit near-term ROI. Don’t pull budget from what’s working — treat this as an additive test, not a pivot.
How do I measure ROI from ChatGPT ads?
This is one of the harder questions in early-stage channel testing, and honest attribution on a new platform is genuinely tricky. We break it down in our dedicated guide on how to measure ROI from ChatGPT ads — including what you can track reliably today and where the gaps still are.
Thinking About Adding ChatGPT Ads to Your Channel Mix?
The worst time to learn a new ad platform is after your competitors have already figured it out. The second worst time is before you’ve pressure-tested your fundamentals on channels that are already proven.
Before you invest time (or money) in ChatGPT advertising, it’s worth knowing whether your existing paid search is running as efficiently as it should. If your Google Ads campaigns have structural issues — poor account organization, unchecked wasted spend, bidding strategy misalignment — adding a new channel won’t fix that. It’ll just give you two underperforming channels instead of one.
If you want an honest read on where your current paid media stands before expanding into new territory, we offer a no-obligation Google Ads account audit that tells you exactly what’s working, what’s not, and what needs to change. No sales pitch wrapped in a “review.” Just a real assessment from people who’ve managed over $50M in spend.