If you've been using ChatGPT on the free tier lately and noticed something new appearing at the bottom of your answers — that's an ad. Not a hallucination. An actual, labeled, sponsored placement.
OpenAI officially started testing ads in ChatGPT on February 9, 2026. For the millions of people who use ChatGPT's free tier as part of their daily workflow, this is a pretty significant shift. I've been watching this rollout since the initial announcement in January, and there's a lot more going on here than the headline suggests.
Let me break down what's actually happening — from how the ads work, to what data OpenAI is using to target you, to whether you can do anything about it.
What changed and when
OpenAI announced the advertising pilot on January 16, 2026. Testing began in the US on February 9, initially reaching about 1% of ChatGPT mobile users. By late March, that number had expanded to around 5% of mobile users, and the rollout has continued since.
The initial ad model was CPM — cost per thousand impressions, priced around $60 CPM at launch. By late April, OpenAI added a CPC (cost per click) model alongside it, with advertiser bids in the $3 to $5 per click range. That CPM rate dropping from $60 to around $25 in a matter of weeks is probably what pushed OpenAI to add the CPC option — basically a sign that demand isn't quite keeping up with inventory yet.
Major ad agency holding companies — WPP, Omnicom, and Dentsu — have already secured placements in the pilot program. This isn't some small experiment. OpenAI is building real advertising infrastructure here, and they're doing it fast.
What the ads actually look like
According to OpenAI's own description, ads appear at the bottom of ChatGPT's answers — below the response, clearly labeled as sponsored, visually separated from the actual content. You're supposed to be able to tell at a glance that you're looking at an ad, not part of the answer.
The placement is contextual. If you ask ChatGPT about, say, project management tools, there's a reasonable chance you'll see an ad for a software product in that category underneath the response.
OpenAI has been firm on one point: ads don't influence ChatGPT's answers themselves. The model isn't being steered toward recommending products that are advertising. The ad sits below the answer, separate. Whether that separation holds up in practice over time is a fair thing to watch skeptically — but for now, that's the stated design and there's no evidence it's working otherwise.
Who sees ads — and who doesn't
This is the most important thing to understand, and it's cleaner than you might expect.
Sees ads:
- Free tier users (US first, expanding globally)
- Go tier users ($8/month)
Does NOT see ads:
- ChatGPT Plus ($20/month)
- ChatGPT Pro ($200/month)
- Business plan
- Enterprise plan
- Education plan
If you're on Plus or above, you're ad-free. Full stop. OpenAI isn't testing ads on paying subscribers.
The Go tier is interesting — it's the $8/month tier that sits between free and Plus. Cheaper than Plus, but apparently not premium enough to escape ads. That's a meaningful product distinction, and it matters if you're considering which tier is actually worth paying for.
The privacy picture — and where it gets complicated
Here's where I want to slow down, because OpenAI's advertising approach involves a few distinct privacy layers that are easy to conflate.
What OpenAI says it DOESN'T do:
- Share your actual chat conversations with advertisers
- Sell your data to advertising partners
What it DOES do:
- Target ads based on the topic of your current conversation, your past chats, and your previous interactions with ads in ChatGPT
- Share limited identifiers — specifically cookie IDs, device IDs, and email addresses — with advertising partners for cross-platform targeting (like showing you OpenAI promotions on Instagram)
And here's the part that flew under the radar a bit: on April 30, 2026, OpenAI quietly updated its US privacy policy and switched marketing cookies on by default for free users. Not opt-in. On by default.
Plus and Enterprise subscribers? Exempt. They weren't switched to default tracking. So there's now a two-tier privacy system operating alongside the two-tier ad system: free users carry the tracking load, paying subscribers don't.
That's worth being clear-eyed about. OpenAI isn't technically selling your conversations — but it is using your conversations to determine what ads to show you, and it's now sharing your device identifiers with marketing partners by default without asking you first. Those are meaningfully different claims, and conflating them flattens something that deserves more attention.
Can you actually opt out?
Short answer: sort of. Here are your real options.
Option 1: Upgrade to Plus or higher. This is the cleanest solution. No ads, no default tracking, done. At $20/month, Plus eliminates ads entirely and keeps you out of the default-tracking tier. If you were on the fence about upgrading, this may tip the decision.
Option 2: Opt out within the free tier. You can turn off personalized ads and marketing tracking on the free tier by going to Settings > Data Controls > Marketing Privacy inside the ChatGPT app or on the website. You can also clear your collected ad data with a single tap from this menu.
The catch: if you opt out of ads entirely while staying on the free tier, you'll get fewer daily free messages. OpenAI is explicitly trading message volume for ad exposure — if you don't see ads, they reduce your daily allowance to compensate. That's a genuine trade-off, not a neutral choice.
Option 3: Use ChatGPT without logging in. Ads currently only apply to logged-in adult users on free and Go tiers. If you're not logged in, you're presumably not being tracked and targeted the same way.
Geographic rollout: where things stand now
The advertising pilot started in the US and has been expanding. The next markets are Canada, Australia, and New Zealand. After that, the EU.
The EU expansion is the one to watch from a privacy perspective. OpenAI has publicly committed to a "consent-first" approach for European users — shaped by GDPR requirements. What that looks like in practice (explicit opt-in consent dialogs vs. softer consent mechanisms) will tell us a lot about how much of the "privacy-respecting" framing is actually real versus performative.
The US launch didn't include a meaningful opt-in flow. Marketing cookies were just switched on by default. If the EU launch looks significantly different, that's an implicit admission that the US rollout could have been handled better.
What this means for ChatGPT as a product
I want to be honest about something here: ChatGPT with ads isn't fundamentally broken. The answer quality hasn't changed. The interface still works. For the vast majority of queries, you'll get a useful response and an ad below it that you can ignore.
But it changes what kind of product ChatGPT is.
For three-plus years, one of ChatGPT's compelling value propositions was that it felt like a tool — you talked to it, it helped you think, there was no marketing agenda in the interaction. That's a different psychological contract than "free product supported by advertising." It's the difference between a calculator and an ad-supported app. The calculator doesn't have interests.
That's not necessarily disqualifying — ad-supported products can be useful, and $0 for access to a capable AI assistant is still a pretty good deal. But it's honest to name the shift.
The privacy opt-in-by-default issue bothers me more, frankly. Switching on tracking cookies without proactively notifying free users is the kind of thing that erodes trust quietly. It's easy to miss, and the people most likely to miss it are exactly the people who least understand what's being collected.
If you're a casual free-tier user who doesn't know you're now sharing device identifiers with marketing partners by default, go check your settings. Settings > Data Controls > Marketing Privacy. Takes 30 seconds.
Our take
This was always going to happen. OpenAI needs revenue that isn't just subscription-based, and advertising is the default monetization model for products with large free audiences. The more interesting question is how they execute it over time — and whether the "ads don't influence answers" commitment stays intact under commercial pressure.
For now: if you're using ChatGPT free, you'll start seeing ads if you haven't already, your marketing cookies are probably on by default, and you have limited but real opt-out options. If you're a heavy user, the upgrade math to Plus gets more compelling now — you're not just paying for higher limits, you're paying to stay out of the ad tier entirely.
For a full breakdown of ChatGPT's features, pricing, and where it stands in 2026 more broadly, check our ChatGPT review. And if you're specifically evaluating ChatGPT against other AI tools for technical work, our Claude vs. ChatGPT for coding comparison has details on how the two stack up where it actually matters.
Sources: OpenAI's advertising announcement and privacy policy update (openai.com), MacRumors, SearchEngineLand, Digiday, CNBC.
Top comments (0)