ChatGPT ads policy: which businesses can advertise and which are blocked in Israel

The ChatGPT ads policy is the first thing to check before you open an ad account, even before the budget. ChatGPT Ads opened to Israeli advertisers in mid-September 2026, but the policy largely closes it to fields that Israeli businesses advertise in every day: insurance, mortgages, clinics, lawyers, and ads for a specific flat or job. This article goes through the categories in OpenAI's own wording: what is allowed, what is blocked outside the US, what sits in the gray area and how to check your business. Everything here is correct as of 22 September 2026, based on the policy version of 10 September 2026. The technical side (budget, account setup and availability in Israel) is covered in our guide to advertising on ChatGPT.

Three checks: the advertiser, the ad and the conversation

Under OpenAI's advertising policies, every ad goes through three checks, and any one of them can stop you:

1. The advertiser. During account verification, OpenAI considers whether what you sell is eligible for ads. According to the account setup page, this check is part of the verification process, and your industry is one of the business details you fill in.
2. The ad and the landing page. The headline, copy, media and landing page are reviewed together. An ad is rejected if any of them promotes a disallowed category or meaningfully references one.
3. The conversation. Even an approved ad appears only in conversations that comply with the placement policy. In some conversations, no ad will appear at all.

The key distinction: the first two checks decide who gets to advertise, and the third decides where. So a business in an allowed category may find that its ad does not show in exactly the conversations that looked most relevant to it. Most reviews are automated, using language models and classifiers with human oversight. Restricted categories may require enhanced verification or manual review.

ChatGPT ads policy: the categories allowed today

At this stage, OpenAI names four families of consumer categories as its focus. The policy defines families, not a list of businesses, so the Israeli examples below are ours:

• Consumer and household goods: online shops, fashion, furniture, homeware.
• Local services: plumbers, electricians, pest control, cleaning, removals and renovations.
• Travel, experiences and entertainment: guesthouses (zimmers), hotels, attractions, tours and shows.
• Digital products and education: apps, software and courses.

It is worth being clear about what "focus" means. The policy states explicitly that all other categories are disallowed at launch. So a business that does not clearly fit one of the four families, such as a B2B service or an agency, is in undefined territory, and the decision is made at account verification. OpenAI says it expects to expand the list over time, but has given no date.

The categories blocked for businesses in Israel

This is where the wording is easy to misread. In the introduction to its content rules, OpenAI writes that it may approve advertisers in financial services, healthcare and legal services gradually, manually and case by case. Yet in each of those categories' own sections, the permission covers the US only. Outside the US, finance and health are generally prohibited and legal services are currently prohibited. For an Israeli business the practical upshot is simple: the channel is effectively closed.

Some financial services are not allowed even in the US: credit repair, debt settlement, debt assistance, and alternative investments such as bullion.

On top of that, these are currently disallowed for every advertiser: adult content (including dating apps and sexual health products), alcohol over 0.5% and tobacco (including vaping and nicotine products), gambling (including lotteries), political content, cannabis and psychedelics, and counterfeit goods. Gambling has two narrow exceptions: lodging or entertainment at a casino resort where gambling is not the focus, and digital games with no real-money wagering or cash prizes. Weapons also appear in the policy as an example of a restricted or prohibited category.

One more point that is easy to miss: under the policy, an advertiser whose primary business model falls within a prohibited or restricted category may be ineligible to advertise at all. In other words, a clinic that also sells creams will not necessarily be able to advertise just the creams. If your business is in the table, your budget is better spent in channels that already work, such as Google Ads campaign management.

The gray area: same business, allowed or not depending on the wording

In some categories, what decides is not the type of business but what the ad claims:

• Fitness and nutrition. Fitness equipment and workout programmes may be permitted, as long as no medical claim is made. Health coaching, diet pills and detox programmes, on the other hand, are listed as not permitted. Supplements widely sold through major retailers may also be permitted, again with no medical claims. Note that dietary supplements also appear on the health-services list that OpenAI may approve in the US only, so for an Israeli business this is a gray area.
• Beauty and cosmetics. The policy does not name beauty salons. Minimally invasive cosmetic procedures are listed under health services, so outside the US they are generally prohibited. "Acne treatment" is a health claim too. Health-related products that exploit body-image concerns are not allowed.
• Finance without a financial service. A budgeting app, general financial software, or a course about money that does not offer a financial service is generally permitted.
• Law without a legal service. A podcast or educational materials about law may be allowed, as long as no legal service is offered.
• Real estate and recruitment. You may link to a platform that carries property or job listings, provided the ad and the landing page do not reference a specific listing. The policy does not say whether a single brokerage's website counts as a platform.
• Restaurants and events. An incidental mention of alcohol is acceptable only when alcohol is not the focus and the ad otherwise meets the rules. And an allowed ad cannot lead to a page with disallowed content. OpenAI's example: a food delivery ad that links to alcohol delivery.
• Lingerie and swimwear. May be allowed in a standard fashion and retail context, when not framed to provoke sexual arousal.

Even in an allowed category: conversations where no ad appears

The placement policy defines conversations in which no ads are shown, whoever the advertiser. There are two groups:

• Sensitive conversations: conversations about personal or mental health, emotionally reliant conversations, and "sensitive user journeys", a term OpenAI does not define.
• Inappropriate topics: among others, politics, debated social content, graphic violence, self-harm, gambling and tobacco, privacy, and fraud.

One thing changed in April 2026: medical, legal and financial advice conversations are no longer categorically blocked from ads. Personal and sensitive conversations, however, remain blocked. For example, a general question about home workout equipment can show an ad from a fitness equipment store. A conversation in which someone describes a personal health problem will not.

In addition, according to the user FAQ page, OpenAI does not show ads to accounts identified as belonging to users under 18. If your buyers are teenagers, this channel is not built to reach them.

Rules that apply to every ad, in every category

Even in an allowed category, an ad that breaks the baseline standards is rejected:

• Truthful advertising. No unfounded claims about capabilities, pricing, outcomes, affiliations or comparisons with competitors, and no false endorsements.
• Professional language. Obscene or shocking language may get an ad rejected, even when it appears in a product or event name.
• No imitating ChatGPT. An ad that imitates the look, functionality or voice of ChatGPT so that it seems part of the product may be removed or require changes.
• Sensitive social issues. An ad that meaningfully references issues such as abortion, immigration, race, religion or gender identity is not permitted.
• A landing page that can be reviewed. An ad or page the system cannot review will not run, and a page that blocks OpenAI's crawlers is a common reason for rejection. We set out what makes a good page in our landing page guide.
• Licensing. Anyone offering a service that requires a licence or certification must hold it.

In the September 2026 version, OpenAI clarified its right to decline advertisers, including when they conflict with its advertising principles, business interests or competitive position. The wording does not explain how this is applied, but anyone selling a competing AI product should know the clause exists.

How to check your business before you open an account

Five questions to answer before you invest in creative:

1. What is your primary business model? If it is in a blocked category, even an allowed product you sell will not necessarily get through.
2. Which allowed family do you belong to? If there is no clear answer, the decision will be made at account verification, and rejection is possible.
3. Is there a medical, financial or legal claim? "Relieves back pain" or "guaranteed returns" moves an allowed ad into prohibited territory.
4. Where does the ad lead? The landing page must not introduce disallowed content, even when the ad itself is allowed.
5. Does the ad feature a specific flat or job? If so, it will not be approved.

And if the ad is rejected? According to the troubleshooting page, it shows as Not serving in the Ads tab, and hovering over the ad shows the reason. You can edit it and resubmit, or request another review if you believe the rejection was a mistake. On the account setup page, OpenAI also says it is currently unable to expedite account or ad reviews, so it pays to submit a clean version the first time.

What Simple Web does about it

Simple Web is an AI-first marketing agency from Bnei Brak, certified Meta partners and Google advertising experts, working with 200+ businesses and holding 28 five-star Google reviews. For businesses in categories that ChatGPT Ads keeps closed for now, the paid channels that already work remain the core. We run campaigns in Google Ads and on social media, with conversion measurement and a monthly cost-per-lead report. Paid-media management starts at ₪3,500 a month, not including the ad budget, and the ad accounts always stay yours.

For businesses in allowed categories, we also run ChatGPT ad campaigns: we check the category against the policy before the account is opened, write the ads and context hints, and install conversion tracking.

Keep in mind that the ad policy applies to ads only. Appearing inside ChatGPT's answers is a separate, organic channel, which we cover in our guide to AI search optimisation (GEO). Our free marketing audit checks, among other things, your visibility in ChatGPT, Perplexity and Gemini, and the report arrives in Hebrew on WhatsApp within 48 hours.

Summary

The ChatGPT ads policy opens the channel mainly to consumer goods, local services, travel, and digital products and education. Outside the US, finance and health are generally prohibited and legal services are currently prohibited, while ads for a specific flat or job are closed to everyone. In between lies a wide gray area, where the wording of the ad and the landing page decides. The policy has gone through seven versions since March 2026, so check it again before every campaign. Want to know whether your business can advertise on ChatGPT? Book a free diagnostic call.

Sources

• OpenAI: advertising policies: the allowed, restricted and prohibited categories, the placement policy and the changelog (version 1.6, updated 10 September 2026)
• OpenAI: Ads Manager account setup: the ad-eligibility check during account verification, and that reviews cannot be expedited
• OpenAI: troubleshooting common issues in Ads Manager: what to do when an ad is rejected
• OpenAI: ads in ChatGPT, user FAQ: no ads near sensitive topics, and no ads for accounts identified as belonging to users under 18