OpenAI’s new Ads Manager Beta matches ads to the intent of a conversation instead of keywords or demographics, the first ad platform built for chat rather than search results or web pages. The workflow feels familiar to anyone who has run Google or LinkedIn ads, but it lacks audience forecasting, and ads only appear for Free and Go users, not the paying Plus, Pro, Business, and Enterprise power users most B2B brands want to reach. Privacy separation between ads and ChatGPT’s answers is the whole ballgame. The real question is whether OpenAI can bring paid subscribers into the ad pool without breaking the trust that made ChatGPT work.
I’ve spent the past few weeks digging into OpenAI’s new Ads Manager Beta. My first instinct was to compare it to Google Ads, LinkedIn, and Meta. Instead I ended up asking a different question.
Who exactly are advertisers paying to reach?
The answer says a lot about both the promise and the current limits of AI-native advertising.
Advertising Without Keywords
The most interesting part of this platform isn’t where the ads show up. It’s how they get picked.
Traditional digital advertising runs on keywords, interests, and demographics. Google asks what you searched for. Meta asks who you are. LinkedIn asks where you work.
OpenAI asks something different.
Advertisers describe the kinds of conversations where their product might be relevant, using what OpenAI calls “context hints.” The platform reads the intent of the conversation and decides whether an ad fits. It’s semantic matching, not keyword matching, and the ads stay clearly separated from ChatGPT’s answers.
This is the first advertising platform I’ve seen built for conversations instead of web pages or search results.
A Surprisingly Familiar Setup
The targeting model is new. Most of the rest of the platform is not.
Campaigns have ad groups and ads. Advertisers pick CPM or CPC pricing, set budgets, upload creative, and track impressions, clicks, CTR, and conversions once a campaign is live.
If you’ve run Google Ads or LinkedIn campaigns before, you’ll recognize the workflow immediately.
Then you notice something missing.
Where’s the Audience Planner?
Most advertising platforms answer one question before you spend a dollar: how many people can I actually reach?
Google estimates impressions. Meta forecasts audience size. LinkedIn estimates professional reach.
OpenAI doesn’t. Not yet, anyway.
Right now advertisers pick countries, set budgets, write context hints, submit for review, and wait to see what happens. There’s no audience forecasting tool, no reach estimate, no visibility into available inventory before the campaign starts serving.
If you’re used to detailed media planning, that’s a real gap.
The Elephant in the Room
There’s a bigger question, and it might matter more than the missing forecasting tools.
Ads only show up for users on the Free and Go plans. Plus, Pro, Business, Enterprise, and Education users don’t see them. Neither do users under 18.
That’s an odd setup when you think about who’s actually using ChatGPT the most.
Software developers, executives, consultants, researchers, the power users who’ve built ChatGPT into their daily workflow. Those are exactly the people most likely to be paying for Plus or Pro already. They’re also exactly the audience most B2B tech companies want to reach.
Instead, today’s ad inventory sits with the free and low-cost tier.
That doesn’t make it worthless. For a lot of consumer brands, reaching millions of everyday users while they’re actually researching a purchase could work extremely well.
But it leaves an obvious question on the table: how much value is OpenAI giving up by keeping paid subscribers out of the ad pool entirely?
Privacy First, at Least
Credit where it’s due. OpenAI has been clear on one thing.
Advertising is kept separate from ChatGPT’s actual responses. Advertisers don’t get conversation transcripts or personal data, and ad selection happens independently of what the model says back to you. Users get controls over ad personalization too.
That separation has to hold if conversational AI is going to keep people’s trust. I don’t think that’s a nice-to-have here. It’s the whole ballgame.
What I Actually Want to Know
I like poking at platforms like this, and there’s a list of questions I can’t answer yet because the tools aren’t there:
Which conversation topics generate the most inventory? Which industries are fighting hardest for placement? What does effective CPM actually look like? Which context hints perform? How does inventory in Canada compare to the US?
Right now the only way to answer most of these is to run real campaigns and watch what happens.
That’s a fun opportunity if you’re an early adopter. It also tells you how early this platform still is.
The Bigger Picture
I don’t think OpenAI is trying to rebuild Google Ads.
They’re trying to build something genuinely different: an ad platform built on conversations instead of keywords. That’s a much harder problem than the one Google solved twenty years ago.
Before you get sophisticated forecasting and optimization, OpenAI has to learn how conversational advertising actually behaves at scale. The current beta reflects exactly that. Capable for a first release, but missing a lot of the planning tools marketers already expect from everyone else.
The real question isn’t whether conversational advertising works. It’s whether advertisers will settle for reaching only the free tier, or whether OpenAI finds a way to bring paying subscribers into the ad ecosystem without breaking the trust that made ChatGPT work in the first place.
I think that decision, whichever way it goes, ends up being one of the more important product calls in AI this year.
Frequently Asked Questions
How does OpenAI’s advertising platform target ads?
It uses semantic matching rather than keywords, interests, or demographics. Advertisers write “context hints” that describe the kinds of conversations where their product is relevant, and the platform reads the intent of a ChatGPT conversation to decide whether an ad fits. The ads stay clearly separated from ChatGPT’s answers, making it the first major ad platform built for conversations instead of search results or web pages.
Who sees ads in ChatGPT?
Only users on the Free and Go plans, and not anyone under 18. Plus, Pro, Business, Enterprise, and Education subscribers do not see ads. That is a notable limitation, because the paying power users excluded from the ad pool, including many developers, executives, and consultants, are exactly the B2B audience many advertisers most want to reach.
What is missing from the OpenAI Ads Manager Beta?
The biggest gap is audience planning. Unlike Google, Meta, and LinkedIn, OpenAI offers no reach estimate, no audience forecasting, and no visibility into available inventory before a campaign runs. Advertisers pick countries, set budgets, write context hints, submit for review, and wait to see what happens. The campaign structure and CPM or CPC pricing feel familiar, but the pre-launch planning tools marketers expect are not there yet.
Is advertising in ChatGPT private?
OpenAI keeps advertising separate from ChatGPT’s actual responses. Advertisers do not receive conversation transcripts or personal data, ad selection happens independently of what the model says back to you, and users get controls over ad personalization. That separation is essential, not optional, because conversational AI only keeps working if people continue to trust it.
Should marketers use OpenAI’s ad platform now?
It depends on your goals. For consumer brands wanting to reach millions of everyday users while they research purchases, it can already work well, and early adopters gain a chance to learn how conversational advertising behaves before competitors do. For B2B advertisers targeting senior professionals, the value is limited today because those users are mostly on paid tiers that do not show ads. The honest answer is that this is an early beta, and the only way to answer most performance questions right now is to run real campaigns and measure.