ChatGPT Sells Ads Now. But Not Next to Your Patients' Health Questions.

Quick answer: ChatGPT sells ads now. OpenAI says the ads business hit a $1 billion annualized run rate in under 200 days — that's extrapolated monthly revenue, not cash in the bank, but it's real and it's growing. Here's what most coverage missed: ads never appear next to personal health conversations. OpenAI says so in writing. And on the advertiser side, pure-play healthcare businesses "may be ineligible" to advertise at all. So the answer a patient gets to a health question is organic only. No auction. No bidding. Just data — accurate or not.

• ChatGPT ads started testing in the US on February 9, 2026, for Free and Go plans. Paid tiers stay ad-free
• OpenAI reports a $1B annualized run rate and 40+ countries. The run rate number extrapolates current monthly revenue — it's not total revenue collected
• Ads sit at the bottom of answers, clearly labeled. OpenAI says they don't influence what ChatGPT tells you — different systems, advertisers can't touch the responses
• OpenAI's policy: ads are "not eligible to appear near sensitive topics, including personal health, mental health, or politics." That's current, updated September 2026
• US dental, vision, and minimally-invasive-cosmetic advertisers can get approved case-by-case. But OpenAI also says advertisers whose "primary business model" is healthcare "may be ineligible"
• Anthropic won't touch ads in Claude at all. The two companies are arguing about this in public. Nobody's won

What's actually running

OpenAI started testing ChatGPT ads on February 9, 2026 — logged-in adults on the Free and Go plans, US first. Ads show up below the response, labeled as sponsored, visually separated from the answer. Plus, Pro, Business, Enterprise and Edu don't see them.

By August 31, OpenAI was reporting a $1 billion annualized run rate — under 200 days after launch. Tens of thousands of advertisers. 40+ countries. Self-serve Ads Manager now live across India, Europe, the Middle East and North Africa.

How ads get picked: a relevance-weighted second-price auction, using conversation context, landing page, ad copy, and — if you've left personalization on — signals from your broader ChatGPT history.

The commitment, repeated across every OpenAI document on this: "Ads do not influence the answers ChatGPT gives you." Separate systems. Advertisers can't shape, rank, or alter responses.

The part that matters for clinics

From OpenAI's live FAQ, word for word:

Will ads appear in sensitive chats? No. Ads are not eligible to appear near sensitive topics, including personal health, mental health, or politics.

The ad policies (last updated September 10, 2026) list "mental and personal health conversations" as sensitive user contexts. Placement there is prevented.

There's a wrinkle here nobody's writing about. The same policy changelog notes that since April 2026, "medical, legal, and financial advice contexts are no longer categorically blocked from ads by default." So general conversations touching medical topics — not an automatic no anymore. But the next sentence of that changelog: "Sensitive conversations and other prohibited contexts remain ineligible for ads." Personal health conversations are still off-limits. The line moved for advice. It did not move for patients.

Now the advertiser side, which is tighter than launch coverage suggested. The health services policy says health ads outside the US are generally prohibited. In the US, OpenAI "may allow ads from approved advertisers on a case-by-case basis" for: dental services, vision products, minimally invasive cosmetic procedures, health insurance, hospitals and urgent care, medical testing and diagnostics, consumer devices and apps, and dietary supplements. Invasive or high-risk procedures: disallowed, no exceptions listed.

And then this line, which should end every "let's run ChatGPT ads for your clinic" pitch:

Advertisers whose primary business model or offerings fall within restricted or prohibited categories (e.g., healthcare, weapons) may be ineligible to advertise on our platform.

So a dental group that also sells whitening kits might get through review. A fertility clinic asking where to buy ads next to fertility questions has two problems: those ads never run next to those questions, and the practice itself may not be an eligible advertiser to begin with.

Put together: when a patient asks ChatGPT "does my biopsy result look concerning" or "is this IVF protocol normal" — there is no ad slot. None. The answer is organic, end to end. Nobody can outbid you for it. Nobody can buy the position above you. What shows up is whatever the data says about you. Accurate, or not.

Why Anthropic said no to all of this

While OpenAI was scaling ads, Anthropic went the other way completely — Super Bowl ad on February 8, 2026: "Ads are coming to AI. But not to Claude."

Their reasoning is worth reading yourself, but two lines:

"Users shouldn't have to second-guess whether an AI is genuinely helping them or subtly steering the conversation towards something monetizable."

"The history of ad-supported products suggests that advertising incentives, once introduced, tend to expand over time as they become integrated into revenue targets and product development, blurring boundaries that were once more clear-cut."

That's not a take from a blog. That's one of the two biggest AI labs saying, officially, that paid incentives tend to bleed into organic results over time.

The argument nobody settled

Sam Altman called that Super Bowl ad "deceptive" and "clearly dishonest" — said OpenAI would "obviously never run ads in the way Anthropic depicts them." Anthropic's commercial chief told CNBC ads put a company in "directions where you're optimizing for the wrong things," and that Anthropic is "not fighting another partner for eyeballs or for ad revenue."

Both positions are on the record. Neither has been tested long enough to be proven. The timeline here is eight months. Five years from now this argument might look silly in one direction or the other. Today, the only thing you can plan around is the written policy: ads stay out of personal health conversations, and healthcare advertisers get extra scrutiny.

What to do with this this week

• If someone pitches you "ChatGPT ads for your clinic," ask two questions. Where exactly would that ad appear when a patient asks about a procedure? (Nowhere, by policy.) And is a pure-play healthcare practice even an eligible advertiser? (May not be, per policy.) Watch what they do with those answers
• Check Claude first when you audit your AI visibility. No ads, ever. Whatever it says about your city and your procedure is pure organic signal. If it names your competitor, that's not an algorithm having a bad day — that's the data
• Directory work still decides everything. The Copilot Health find-a-doctor search, ChatGPT's general answers, Perplexity's citations — they all read from the same kinds of sources. Our five health AIs breakdown covers who can surface a clinic and who can't
• Run the 20-minute check. The visibility walkthrough hasn't changed. The ads news doesn't change the method — it just removes the reason to keep waiting

One honest caveat

OpenAI says ads don't influence answers. Their policies say health conversations are off-limits. Both are current facts. Both could change — the policy's been revised six times since March. Anthropic's "incentives expand" argument is a risk thesis, not documented harm. We're stating mechanisms and documented positions here, not predictions. If the health carve-out ever gets relaxed, everything above gets more important, not less.

Frequently Asked Questions

Can my clinic buy ChatGPT ads?

Maybe — but not the way most agencies describe it. OpenAI's published policy allows US dental, vision, and minimally invasive cosmetic advertisers case-by-case, plus insurance, hospitals, urgent care, and diagnostics. Outside the US, health services ads are generally prohibited. And OpenAI also states advertisers whose primary business is healthcare "may be ineligible." Even if approved, your ad never appears next to a personal health conversation.

Does ChatGPT show ads next to health questions?

No. OpenAI's documentation says ads are "not eligible to appear near sensitive topics, including personal health, mental health, or politics," and the ad policies list "mental and personal health conversations" as sensitive contexts where placement is prevented.

Do ChatGPT ads influence which clinics get recommended?

No — per OpenAI, ads run on separate systems and advertisers cannot shape, rank, or alter ChatGPT's responses. Ads appear labeled at the bottom of answers. When a user asks a personal health question, no ads appear at all, by policy.

Will ChatGPT ads change organic recommendations over time?

Nobody knows. Anthropic has officially argued that ad incentives "tend to expand over time… blurring boundaries that were once more clear-cut." OpenAI has officially committed to answers staying independent of ads. Both are on record; neither has been tested long enough. Plan around today's policies, not predictions.

Why doesn't Claude have ads?

Anthropic says conversations with Claude are often sensitive, personal, or high-stakes — places where ads would feel incongruous. Claude's revenue comes from enterprise contracts and subscriptions. Their commercial chief has said the company is "not fighting another partner for eyeballs or for ad revenue."

If clinics can't buy their way into a health answer, how do AI tools decide which ones to show?

Organic signals: accurate, consistent directory listings; insurance data that matches patient filters; structured content every AI can parse; entity signals that agree across your site, Google Business Profile, Healthgrades, Zocdoc, and specialty directories. The full mechanism is in our five health AIs breakdown.

How do I check what AI says about my practice today?

Ask ChatGPT, Claude, Gemini, and Perplexity the questions your patients actually ask — with your city and procedure in them — and see who gets named. Claude is the cleanest test: no ads, ever, so what it says is pure organic signal. Our visibility walkthrough has the exact prompts.

Curious where your clinic stands? Claim Free Snapshot — I'll check how ChatGPT currently describes your practice before anyone talks retainers, or see how the audit works first.

The clinic that can be cited is worth more than the clinic that can be bid for.