Can My Industry Advertise on ChatGPT? Restricted Categories and What Manual Review Actually Does
By The InPromptAds Team 7 min read
Last verified 15 August 2026. Version 1.0. Next scheduled review 15 September 2026.
Every guide in this category republishes OpenAI's category lists. None of them tells you what happens after you submit in a category that requires review, because that information sits inside advertiser accounts and nobody with one has published it.
This page does two things. It states the rules accurately, including the sentence most summaries drop. Then it is explicit about the size of what is not known, because a vertical eligibility page that pretends the review process is documented would be making the same mistake as the guides it replaces.
The three tiers
Ad policies v1.3, updated 15 July 2026.
Allowed
Advertising is limited to consumer categories: household and consumer goods, local services, travel and experiences, digital products and education.
Read that list twice if you sell B2B. It is a consumer list. There is no enterprise software category on it, and the abundance of "ChatGPT ads for B2B SaaS" content in this market is not derived from the policy.
Disallowed outright
Gambling, alcohol, tobacco, dating and adult content, recreational drugs, counterfeit goods, political content, legal services.
Legal services are not permitted in any market. Legal education is, which is how LSAT prep can advertise and a personal injury firm cannot. That distinction is the single most misunderstood line in the policy.
Case by case, with proof of licensure
Financial and health advertisers may be approved individually, across a defined list covering insurance, mortgages, credit cards, investment services, dental, medical devices and hospitals.
Then the sentence almost every summary drops: ads for financial services outside the United States are generally prohibited, and the same applies to health services.
The case-by-case route is a US carve-out. A licensed insurer in Toronto, London or Mumbai does not get to apply. This is the difference between "regulated but possible" and "not available to you," and it is buried in a clause.
The layer that has nothing to do with your category
A separate rule governs where ads appear regardless of who the advertiser is.
No ads run beside conversations about personal health, mental health or politics, or in what the policy calls sensitive user journeys.
So a permitted advertiser can still find their available inventory much thinner than their category suggests. A supplements brand in an allowed consumer category is selling into a conversation that frequently turns to personal health, at which point the placement disappears. Bidding does not recover it.
This is worth modelling before you commit budget: your inventory is not your category, it is your category minus the conversations your product is actually discussed in. For some products those overlap almost entirely.
What nobody has published about manual review
This is the gap, and we would rather name it precisely than fill it.
| Question | Published anywhere? |
|---|---|
| Approval rate for financial advertisers | No |
| Approval rate for health advertisers | No |
| Median or typical review timeline by vertical | No |
| What documentation actually satisfies "proof of licensure" | No |
| Common rejection reasons within reviewed categories | No |
| Whether a rejection is appealable, and the success rate | Re-review can be requested. Success rate not published |
| Whether review outcomes differ by market inside the US carve-out | No |
Not one of the agencies publishing eligibility guides has disclosed a single approval outcome, and several of them hold live accounts. OpenAI publishes the policy and not the practice.
What OpenAI does say about timing generally: applications sit in a rolling queue with high volume, review may take some time, and OpenAI states it is unable to expedite account review or ad review requests at this time. That is a statement about the queue, not about vertical review.
What to do if you are in a reviewed category
Check the geography clause first, and check it before anything else. If you are a financial or health advertiser outside the United States, the case-by-case route is not open and nothing downstream matters. Ten minutes, free.
Do not submit twice. Duplicate applications do not bypass review, create confusion about which account is real, and can slow the thing you were trying to accelerate. The same applies to inconsistent documents at the verification step.
Get licensure documentation consistent before you start. Verification runs through Persona and compares what you submit against your registered entity. A licence in one legal name and an entity registered in another is a rejection with a slow feedback loop.
Budget the timeline as unknown rather than as a number. No published review timeline exists for reviewed categories and OpenAI will not expedite. A launch date committed to internally on the basis of a blog post's estimate is a date built on nothing.
Publish what happens to you. This category has an information vacuum that only advertisers can fill. If you go through review in a regulated vertical, the outcome and the timeline are worth more to the market than another eligibility summary, and we will cite anyone who publishes theirs.
What we cannot tell you
Anything about review outcomes. We do not hold an advertiser account, we have not submitted an application, and we are not going to model an approval rate from nothing. The table above is the honest state of public knowledge.
Whether the policy's approved-category list expands. Ad policies moved to v1.3 in July. It is a versioned document, which usually means more versions.
How "sensitive user journeys" is operationalised. The policy names the concept. It does not define the boundary, and the boundary is what determines your available inventory.
Whether crypto, supplements, firearms or CBD fall inside any allowed category. None appears on either published list by name. Absence from the disallowed list is not permission, and we would treat anything unlisted as requiring a direct answer from OpenAI rather than an inference.
Quick answers
Can financial services advertise on ChatGPT? In the United States, case by case with proof of licensure, across a defined list including insurance, mortgages, credit cards and investment services. Outside the United States, ads for financial services are generally prohibited.
Can healthcare companies advertise on ChatGPT? Same structure. Case-by-case approval with proof of licensure inside the United States, covering categories including dental, medical devices and hospitals. Health services advertising outside the United States is generally prohibited.
Can law firms advertise on ChatGPT? No. Legal services are not permitted in any market. Legal education is permitted, which is why exam preparation can advertise and a law firm cannot.
What categories are allowed on ChatGPT ads? Household and consumer goods, local services, travel and experiences, digital products and education. The permitted list is consumer-oriented and does not include an enterprise software category.
How long does review take for a regulated category? Not published. OpenAI states applications sit in a rolling queue with high volume, that review may take some time, and that it cannot expedite account or ad review requests.
What is the approval rate for regulated advertisers on ChatGPT? Not published by OpenAI or by any third party, including agencies that hold live accounts.
My category is not on either list. Can I advertise? Absence from the disallowed list is not permission. Categories not named on either list should be confirmed directly with OpenAI rather than inferred.
Sources
| Claim | Source | Tier |
|---|---|---|
| Allowed categories: household and consumer goods, local services, travel and experiences, digital products, education | OpenAI ad policies v1.3, 15 July 2026 | Confirmed, primary |
| Disallowed: gambling, alcohol, tobacco, dating and adult, recreational drugs, counterfeit goods, political content, legal services | Same | Confirmed, primary |
| Financial and health case-by-case with proof of licensure, defined list | Same | Confirmed, primary |
| Financial services advertising outside the US generally prohibited; same for health services | Same | Confirmed, primary |
| No ads beside personal health, mental health or political conversations, or sensitive user journeys | Same | Confirmed, primary |
| Rolling queue, high volume, review may take time, cannot expedite account or ad review | OpenAI Help Center | Confirmed, primary |
| Verification runs through Persona against registered entity details | OpenAI advertiser documentation | Confirmed, primary |
| Re-review can be requested after a rejection | OpenAI Help Center, Troubleshooting | Confirmed, primary |
| Approval rates, review timelines and rejection reasons by vertical | Not published by OpenAI or any third party | Absent |
Related reading
- Who should not buy ChatGPT ads yet — category restriction as one of six disqualifiers
- What has to be true before you can run a ChatGPT ad — the eleven conditions in order
- ChatGPT ads troubleshooting — rejection and re-review mechanics
Changelog
15 August 2026. v1.0 published.
Field kit
Tools
Site
OpenAI (primary)
The InPromptAds Team
Writes about AI-native advertising and the shift from search to conversation.
