QWE AI Academy You ask ChatGPT to help you pick running shoes. It walks you through cushioning, drop, your weekly mileage. Useful answer. And then – right under it, a card with a brand logo and a small “Sponsored” label. That’s the new normal on the Free and Go tiers as of February 2026, and the question most people are asking is the wrong one. It’s not are there ads, it’s which signals are feeding them, and which toggle actually controls that.
This is a tutorial about how ChatGPT serves ads – where they come from, how to read what’s targeting you, and which setting actually moves the needle. Skip the doom takes. The mechanics are knowable, and three of the controls genuinely change the experience.
What just happened (and why the rollout matters this week)
OpenAI began testing ads in ChatGPT on February 9, 2026, for logged-in adult users in the U.S. on the Free and Go subscription tiers. Plus, Pro, Business, Enterprise, and Education tiers do not have ads. The announcement itself landed January 16 – three weeks of “it’s coming” before the switch flipped.
Two months in, the numbers tell you everything about how seriously OpenAI is taking this. According to an Axios scoop reported by Futurism, OpenAI generated $100 million in annual recurring revenue from ChatGPT ads in just two months (as of approximately April 2026). Pricing is moving fast too – CPM rates dropped from $60 at launch to as low as $25 within ten weeks, and OpenAI turned on cost-per-click ads with bids between $3 and $5 per click, per Digiday’s reporting. Translation: the auction is opening up, and what shows up in your chat window is getting cheaper to buy. Expect more ads, more often, with sharper targeting as the year progresses.
The three targeting layers (this is the part most guides skip)
ChatGPT ad targeting isn’t one signal. It’s a stack of three, and you can switch the upper layers off independently. Read this once and the settings panel makes sense.
| Layer | What it uses | Can you turn it off? |
|---|---|---|
| 1. Current conversation | The topic of the chat you’re in right now | No (this is the always-on baseline) |
| 2. Past chat history + ad interactions | Topics from previous chats, ads you clicked or dismissed | Yes – “Personalize ads” toggle |
| 3. ChatGPT Memory | Saved facts about you across sessions | Yes – “Past chats and memory” toggle |
Per OpenAI’s Help Center on ads, the system matches ads with the topic of your conversation, your past chats, and past interactions with ads. If multiple advertisers compete, the most relevant one wins the slot. The catch nobody talks about: accounts with Memory disabled won’t see the past-chats personalization option active – it’s greyed out, confirmed by Search Engine Land and SERoundtable. So if you’ve kept Memory off for general privacy reasons, you’re already running on Layer 1 only.
Setting up your ad controls – the practical 5-minute version
Open ChatGPT, go to Settings, and find Ad Controls (it’s a separate tab from Data Controls – easy to miss). OpenAI’s Help Center lists what you’ll find there: your ads history and interests, a delete button that clears that history and interest profile (with up to 30 days for full removal from systems), a “Personalize ads” toggle, and a “Past chats and memory” toggle that controls whether Memory feeds ad relevance.
Here’s the order I’d recommend if you want to reduce targeting without losing functionality:
- Toggle off “Personalize ads.” This drops you to Layer 1 only – current conversation context. Ads still appear, but they can’t draw on your history.
- Tap “Delete ads data.” This wipes the interest profile that was built before you toggled off. Without this step, the toggle just stops adding new data; old data still shapes your initial sessions.
- Decide on Memory separately. If you use Memory for actual ChatGPT context (your job, your projects), keep it. Just know it’s a third targeting input.
- For sensitive sessions, use Temporary Chat. Per OpenAI’s Help Center, Temporary Chats do not show ads, don’t feed targeting, and auto-delete within 30 days.
The Ads-Free toggle, and the trap inside it
OpenAI gave Free-tier users an unusual option: turn ads off entirely, in settings, without paying. Sounds great. Read the fine print first.
Choosing the no-ads option means fewer messages per day and no access to tools like image generation – confirmed by both OpenAI’s Help Center and independent reporting from Fello AI and TheAutomated. So the Ads-Free toggle costs you image generation and slices your daily message cap. For a casual user who only types text questions, that’s a fair trade. For anyone using DALL-E or regularly hitting the Free cap, it’s a downgrade in disguise.
Then there’s the Go tier confusion. The Go plan costs $8 per month in the US (as of early 2026, per Euronews). A lot of users assumed paying anything would buy them out of ads. It doesn’t. Go is the ad-supported paid tier – you get more usage, you still see ads. The real ad-free threshold is $20/month (Plus and above).
What advertisers actually see (and don’t)
Advertisers don’t see your chats. Full stop. OpenAI’s Help Center is explicit: advertisers have no access to conversations, chat history, memories, or personal details. They receive only aggregated, non-identifying data – total views, total clicks, that sort of thing.
That’s stronger than Meta’s or Google’s data-broker model on paper. The asterisk: OpenAI itself still uses your conversation context to pick which ad to show – the privacy boundary is between OpenAI and the advertiser, not between you and OpenAI. If you’re comfortable with that distinction, the controls work as advertised. If you’re not, the only complete answer is Plus or above.
Honest limitations and unknowns
Ads are not eligible to appear near sensitive or regulated topics, including health, mental health, or politics – that’s the official policy per OpenAI’s Help Center. The unanswered question: how accurate is the classifier that decides a chat counts as “health-adjacent”? OpenAI hasn’t published that number, and a borderline conversation – say, asking about supplements before a run – could plausibly slip through either way. Until OpenAI publishes detection metrics, treat the exclusion as a soft guarantee, not a hard one.
The CPC shift is worth watching separately. CPMs dropping from $60 to $25 within ten weeks, and the move to cost-per-click bidding at $3-$5, signals that inventory is being repriced as advertisers stress-test the platform. Cheaper clicks tend to attract higher-volume, lower-quality bidders – which is a different problem than ad frequency. International rollout timing is also unconfirmed; the test is U.S.-only as of this writing. OpenAI has stated that protecting user privacy will remain central as the program evolves, including guardrails to prevent narrow targeting and continued scrutiny of who gets into the advertiser program. What that means in practice will become clearer over the next few quarters.
For the official source of truth as this changes, bookmark OpenAI’s Help Center page on ads and the testing ads blog post. Both get updated as the pilot expands.
FAQ
Will ChatGPT recommend products for money inside its actual answer?
No. Per OpenAI’s Help Center, ChatGPT responses are generated independently and ads are not integrated into responses. The sponsored unit appears below the answer as a separate, labeled card.
If I turn off personalization, will I see fewer ads or just less relevant ones?
Less relevant, not fewer. OpenAI’s Help Center is clear on this: turn off ads personalization and you still see ads based on the current chat thread – OpenAI just won’t use your other chat history or interest profile to inform them. So if you ask about Italian recipes, you might still get a pasta-brand ad, just not one informed by the fact that you spent three weeks last month researching air fryers. For “fewer ads, period,” the only real lever is the Ads-Free toggle (with its tool trade-off) or upgrading to Plus.
Does any of this apply to me if I’m not in the US?
Not yet. The pilot is US-only as of early 2026. The Ad Controls panel may not even be visible in your region – worth checking your settings every few weeks, since international expansion is the obvious next step and likely won’t come with a separate announcement.
Next action: Open ChatGPT right now, go to Settings → Ad Controls, and do two things: toggle “Personalize ads” off, then tap “Delete ads data.” Takes 20 seconds. If the panel doesn’t exist for your account yet, you’re either outside the test region or on a paid tier where it doesn’t apply – in which case there’s nothing to configure.