Run a clinic, a lending business or a law firm and you have probably read that ChatGPT Ads is closed to you. That is not what OpenAI's policy says. Healthcare, finance and legal are approved-advertiser categories, reviewed manually, one business at a time — a gated path, not a wall. Apply, and run elsewhere while you wait.
OpenAI ad policies, updated June 4, 2026
August 8, 2026
Advertisers in the 7 available countries
That distinction matters more than it looks. "Banned" tells a clinic owner to stop reading. "Gated, with manual review" tells them there is a form to fill in, a wait to plan around, and a decision to make about the next ninety days of budget. Those are different pieces of advice, and only one of them is accurate.
What the policy actually says
OpenAI's ad policies split into two halves that are easy to confuse, and most coverage confuses them.
The first half is the ad content policy — who is allowed to buy ads. During what OpenAI calls the initial test period, that is deliberately narrow:
During the early phases of advertising, we are focused on a limited set of consumer categories, including household and consumer goods, local services, travel and entertainment, and digital products and education.
The second half is the ad placement policy — where ads are allowed to appear, regardless of who bought them. Ads do not run next to sensitive conversations, and OpenAI names mental and personal health conversations and emotionally reliant contexts explicitly.
Here's the thing people miss. In April 2026 OpenAI loosened the placement rule: medical, legal and financial advice contexts are no longer categorically blocked from carrying ads. That change says nothing about whether a clinic can buy them. Two different axes, two different rules, and a lot of write-ups have merged them into one wrong sentence.
Who is genuinely blocked today
Some categories are disallowed outright, not gated. If you are in one of these, there is no approval path to wait for right now:
- Adult content and services, including dating apps
- Alcohol and tobacco, including vaping and nicotine
- Counterfeit goods
- Gambling, including sports betting and lotteries
- Graphic sexual or violent content
- Political content, including advocacy on contested social issues
- Recreational drugs, including cannabis and THC products
- Scams and fraud
- Cryptocurrency, credit repair, debt settlement and debt assistance
- Wellness products making unsubstantiated health claims
Within financial services, those four — crypto, credit repair, debt settlement, debt assistance — are called out as disallowed rather than gated. That is worth reading twice if you work anywhere near consumer credit.
The three categories with a door
Financial services, healthcare and medicine, and legal services sit in a different bucket. OpenAI's wording:
We may approve ads from approved advertisers within the financial services, healthcare & medicine, and legal services categories. These categories are being rolled out gradually with approvals being reviewed manually on a case-by-case basis.
Underneath that headline, each category has a line drawn through it.
| Category | Named as disallowed today | May be permitted |
|---|---|---|
| Healthcare & medicine | Prescription drugs, clinical care providers, hospitals, prescription services, over-the-counter medications, products or services offered through providers or insurance networks | General health and wellness making no medical claims — fitness equipment, wearables, menstrual products |
| Financial services | Cryptocurrency, credit repair, debt settlement, debt assistance. Other financial products and services are restricted to approved advertisers | Financial tools that do not promote financial products or transactions — budgeting apps, general financial software, finance education without offers |
| Legal services | Legal advice, representation and services — including immigration, personal injury, legal claims, document preparation | General legal education or media where no legal services are offered |
Translation: the test is not what industry you are in. It is whether your ad is selling a regulated product or service, or selling something adjacent to one.
What a rejection looks like — and what an appeal did
Policy pages describe the rule. They do not tell you what actually arrives in your inbox. Here is the email OpenAI sends when an advertiser account is turned down:
The reason given is the broadest one on the shelf — "Your business, product, service, or industry is not currently eligible to advertise under our Ad Policies" — followed by the consequence, stated plainly: "Because your advertiser account was not approved, you won't be able to create or run campaigns at this time."
Taken on its own, that is a closed door, and most owners treat it as one. But the email ends with a line that is easy to skim past — "If you have questions or think there has been an error, you can use the button below to initiate an appeal" — above a button marked Initiate appeal.
We clicked it. A few days later, from OpenAI's trust and safety address:
After reviewing your appeal, we determined that we incorrectly restricted your advertiser account from serving ads. We sincerely apologize for any inconvenience this may have caused. Your account's ability to serve ads has been restored in Ads Manager.
Incorrectly restricted. Not "your category has since opened", not "your paperwork finally cleared" — the first decision was wrong and the appeal caught it. This has happened on every account we opened in a gated category, and the turnaround has been days rather than weeks each time. What the appeal took is covered in advertising in a restricted category.
Two things follow from that. The first review is a fast pass that can misread a business, so a rejection naming your entire industry is an opening position rather than a verdict. And an appeal costs nothing except the time it takes to explain what you actually sell — a small price against writing off a channel for a year.
What this means if you run a clinic
"Clinical care providers" is named. A dental practice, a physiotherapy clinic, a medical aesthetics business advertising treatments — all of those are in the gated bucket today, and the approval is manual.
There is a second constraint stacked on top, and it is the one that surprises people. Even with an approved advertiser account, the placement rule keeps ads away from mental and personal health conversations. A meaningful share of the conversations where a clinic would most want to appear are ones where ads do not run at all. Approval opens the door; it does not put you in every room.
What this means if you finance or sell vehicles
Split the question. Selling cars is retail. Financing them is a financial service, and if the offer involves credit repair, debt settlement or debt assistance it is on the disallowed list rather than the gated one.
The practical consequence for an auto business is that the same landing page can decide the answer. OpenAI reviews the whole chain:
These standards apply to all ad assets, including copy, images or video, and landing pages. Ads must be consistent end-to-end.
An ad about inventory that lands on a finance-application page is a finance ad. If you are going to apply, split the campaign from the credit offer first, or the review will do it for you.
What to run instead, while you wait
This is the part that actually moves money, and it is not complicated. If a channel will not take your business today, put the budget where buyers already are and revisit in a quarter.
For local, urgent categories — clinics especially — search intent is still the strongest signal available anywhere. Somebody typing a treatment and a city into Google is closer to booking than almost any conversational query, and that has not changed because a new channel launched. Local Services placements sit even closer to the booking, where the category qualifies for them. Watch your cost per lead rather than your click volume, and the comparison stays honest.
For considered purchases with a longer window — professional services, vehicles, anything with a comparison phase — paid social carries the offer better than search does, because you are creating the demand rather than catching it.
If you want the arithmetic on how the two channels differ before you plan around either, the ChatGPT Ads and Google Ads comparison covers what you can control and what a click costs on each.
The point is that "we can't advertise on ChatGPT yet" is a scheduling problem, not a marketing problem. I have never seen a business fail because it was absent from a three-month-old channel. I have watched plenty burn money on channels they were present on and not watching.
How to keep up when the policy moves
This page quotes the June 4, 2026 version of the policy. It will change — OpenAI says so directly, telling advertisers it expects to expand eligible categories "as our safeguards, review systems, and compliance infrastructure mature."
Two habits are worth building. Re-read the policy page itself rather than a summary of it, because summaries are where the placement rule and the advertiser rule get merged. And re-check before each campaign build, not on a calendar — the categories most likely to open next are exactly the ones currently gated.
If you have decided the wait is worth it, the practical next step is how to advertise on ChatGPT in a restricted category — what approval reviews, how to structure an offer that is not restricted in the first place, and the end-to-end rule that fails more campaigns than the category rules do.
Common questions
Is healthcare banned from ChatGPT Ads? No. It is a gated category, reviewed manually per advertiser. Specific things inside it are disallowed — prescription drugs, clinical care providers, hospitals — and general health and wellness advertising that makes no medical claims may already be permitted.
How long does approval take? OpenAI does not publish a timeline for restricted-category review. It states that restricted categories may require enhanced advertiser verification or manual review, on top of the standard checks. Plan for it to be slower than a normal account approval. When we have appealed a rejection, the reply has come back within days — but that is our experience, not a service level anyone has promised.
My account was rejected. Is that final? Not necessarily. The rejection email carries an "Initiate appeal" button, and when we used it OpenAI came back saying it had incorrectly restricted the account and had restored its ability to serve ads. A first-pass rejection naming your industry is worth contesting before you accept it — the emails are above.
Can I run a general brand ad instead of a treatment ad? That is the right instinct, and it is also exactly what the end-to-end consistency rule tests. The creative and the landing page are reviewed together. If the destination sells the regulated thing, the ad sells the regulated thing.
My category is allowed but my ad was rejected. Why? Baseline standards apply regardless of category — truthfulness, professional language, no imitation of the ChatGPT interface, and a landing page that matches what the ad depicted. An allowed category is necessary, not sufficient.
Does any of this affect my Google or Meta campaigns? No. This is OpenAI's policy for OpenAI's platform. Every network runs its own verification for regulated categories, and being gated on one says nothing about the others.
Is it worth waiting for at all? That depends on what a click is worth to you, which is why we publish what we actually pay across our ChatGPT accounts rather than quoting OpenAI's recommended bid back at you. Start there, then decide whether the queue is worth joining. The broader picture is on the ChatGPT Ads for small business page.
Not sure which channels will take your business?
Bytown's managed service works out where a regulated or gated business should actually be spending, and runs it. Real marketers, one flat fee, full account transparency.
Talk to a marketer Already cleared to advertise? See how to set up a ChatGPT campaign.