What you can advertise
During the initial test, OpenAI says ads are mainly limited to consumer categories: lifestyle and household goods, local services, travel and experiences, and digital products or education1. Every other category is disallowed at launch, though OpenAI says it may approve some advertisers in three restricted categories1.
Restricted: finance, health and legal
OpenAI may approve ads from approved advertisers in financial services, healthcare and medicine, and legal services. It says these categories are rolling out gradually, with each approval reviewed manually1. Within them, the detailed rules are strict:
| Category | Not allowed | May be allowed |
|---|---|---|
| Financial services | Cryptocurrency, credit repair, debt settlement and debt assistance | Approved financial advertisers; budgeting apps, financial software and finance education that don't offer financial services1 |
| Healthcare and medicine | Regulated medical products, services or claims, including prescription drugs, clinical care, hospitals and over-the-counter medicines | General health and wellness products with no medical claims, such as fitness equipment and wearables1 |
| Legal services | Legal advice, representation or services for individuals or businesses, including immigration, personal injury and document preparation | General legal education or media, such as legal podcasts or LSAT prep courses1 |
What's disallowed
These categories are currently disallowed outright1:
- Gambling: anything that involves wagering money on games of chance, skill or uncertain outcomes1.
- Alcohol and tobacco: including drinks over 0.5% ABV, cigarettes, vaping and nicotine products1.
- Political ads: ads for or against political actors, elections, public policy or contested social issues1.
- Dating and adult content: including sexual content, products and services1.
- Drugs: listed with alcohol among the categories disallowed at launch1.
- Deceptive ads: OpenAI prohibits ads that deceive, defraud or exploit users1.
Every ad must also be truthful and clearly distinguishable from ChatGPT itself, and its landing page has to match what the ad shows1.
Where ads never appear
The rules cover the conversation as well as the ad. OpenAI says it won't show ads in conversations about sensitive or brand-unsafe topics, including1:
- Child safety, suicide and self-harm
- Graphic violence, terrorism and weapons
- Hate and harassment
- Fraud, misinformation and privacy
- Political content and debated social issues
- Regulated goods such as gambling or tobacco
Ads are also kept out of emotionally reliant conversations, mental and personal health conversations, and other sensitive user journeys1. In April 2026, OpenAI stopped blocking medical, legal and financial advice conversations from ads by default, and now applies a more precise approach to them1.
How ads are reviewed
Most reviews run through AI systems, including LLMs and classifiers, that check creatives, landing pages and advertiser signals before an ad is approved. People review borderline cases, complex policy areas and high-severity violations1.
If an ad breaks the rules, OpenAI can reject or remove it, limit its delivery, require edits or restrict the advertiser's access. Severe, repeated or deceptive violations can lead to suspension or termination of the account, and higher-severity ads can be paused during an investigation1.
Before you submit
This is our own guidance, based on the policy above.
- Check that your product sits in an open category. If it's financial, health or legal, expect a manual approval.
- Remove medical or health claims from the copy, the images and the landing page.
- Make sure the landing page sells exactly what the ad shows.
- Don't design creative that looks like ChatGPT's own interface.