The ad unit Who can buy What it costs Measurement Notes FAQ Book an intro call

QueryBid/Notes

Eligibility

Which categories can actually advertise on ChatGPT

The eligible list is much shorter than the coverage suggests, three categories sit behind manual approval, and one is closed outright. Here is the whole thing, with the distinction that keeps costing advertisers money.

By Will Tjernlund Published Updated 9 min read

Four categories can buy today: household and consumer goods, local services, travel and experiences, and digital products and education. Financial services and healthcare sit behind manual, advertiser-by-advertiser approval and are limited to the United States. Legal services are prohibited. Everything else is disallowed for now.

Almost every guide to this channel ranks categories by how well they are likely to perform. That is a useful exercise, and it is also the second question. The first is whether OpenAI will accept your money at all, and during the current test period the answer is no for a surprising number of businesses.

We check this before every launch and again before every renewal, because the rules have moved four times since March 2026 and the changes arrive as quiet edits to a policy changelog rather than announcements. What follows is the state of things as of the date on this post.

The eligible list

These four categories are open to any verified advertiser in a supported country, with no special review beyond standard creative and landing page checks.

  • Household and consumer goods. The core permitted vertical and the one the ad format was designed around. The product carousel, which pulls multiple items from a retailer feed, is aimed squarely here.
  • Local services. A strong structural fit. People ask cost and how-to questions first and decide to hire second, and region, DMA, and postal targeting map cleanly onto a service area.
  • Travel and experiences. Long, descriptive planning questions with a real budget attached. Among the best matches on the platform.
  • Digital products and education. Comparison cycles that run almost entirely through research questions, where creative matters more than bid.

If you are in one of these four, you can be live this month. The gate is not your obstacle and your attention belongs on tracking and creative instead.

The restricted list

Two categories were moved into controlled territory by a policy update on July 15, 2026. They are not open, and they are not closed. They are conditional.

Financial services

Approved advertisers only, cleared one at a time through manual review, with enhanced verification and credential requirements, and limited to the United States market. Several subcategories remain prohibited outright, including credit repair, debt settlement, and alternative investments.

Healthcare and medicine

The same structure: approved advertisers, manual review, credential proof, United States only. The update named areas that may qualify, including health insurance, dental, vision, supplements, testing, and hospitals. Pharmaceutical and clinical treatment claims are not on the table, and ads still never serve into personal health conversations regardless of who the advertiser is.

The practical consequence for both: nobody can promise you a launch date. An agency that quotes you a start date for a restricted category before approval has been granted is quoting you a hope. The honest version is that we prepare the application and the substantiation, submit it, and tell you what we hear.

The prohibited list

Legal services are the significant one, because the demand is enormous and the confusion is widespread.

OpenAI's ad content policy states that ads for legal advice, representation, or legal services are not permitted. The policy names the subcategories directly: immigration, personal injury, legal claims, and document preparation. The single narrow opening is general legal education that offers no services, and because OpenAI reviews the landing page as well as the creative, routing an educational ad to an intake form does not survive review.

Beyond legal, the launch period simply excludes what it does not name. Categories outside the eligible and restricted lists are disallowed for now, which catches a great many legitimate businesses that would perform well if they were permitted.

The distinction that costs people money

Placement eligibility and advertiser eligibility are two different things. In April 2026, OpenAI stopped categorically blocking ads from appearing near medical, legal, and financial advice conversations. That governs where any advertiser's ad may show. It did not make law firms, clinics, or lenders eligible to be the advertiser.

Both facts are true at the same time, which is exactly why they get conflated. A travel advertiser's card can appear beneath a response to someone describing a car accident. The personal injury firm still cannot buy that slot.

We have watched this become a sales tactic. A firm reads coverage of the April change, hears that ChatGPT now allows ads in legal conversations, and signs a retainer against inventory it is not permitted to buy. If someone is pitching your firm a ChatGPT ad buy today, the question that settles it is short: ask to see the written approval.

What to do if you are not eligible

Say so and stop, is the first answer. This channel is one line item, it is months old, and being locked out of it this quarter is not a strategic emergency.

The second answer is that the underlying behavior is real even where the ad inventory is not. People are describing their problems to an assistant and acting on what it says. You cannot buy that placement, but you can influence whether a model has anything accurate to say about you, and that work does not depend on anyone's ad policy. It is a slower, less measurable program than paid media and it should be described that way rather than sold as a substitute.

The third answer is to check again in a quarter. This list has changed four times in six months, and the direction of travel has been toward opening categories rather than closing them.

How we verify this. Every platform claim on this page is checked against OpenAI's published ad policies and advertiser documentation, and re-checked monthly. Category rules and pricing in this channel change frequently, and figures reported by early advertisers are not official rate cards. Last verified . Source of record: OpenAI ad policies. If something here has gone stale, tell us at hello@querybid.com and we will fix it.

Want to know where your category stands? We will check it and tell you in a business day.

Check my category

Quick answers

The short versions

Which categories can advertise on ChatGPT?

During the current test period OpenAI permits household and consumer goods, local services, travel and experiences, and digital products and education. Financial services and healthcare are restricted to approved advertisers cleared through manual case-by-case review and limited to the United States. Legal services are not permitted. Categories outside these lists are disallowed at launch.

Can law firms advertise on ChatGPT?

No. OpenAI's ad content policy states that ads for legal advice, representation, or legal services are not permitted, naming immigration, personal injury, legal claims, and document preparation directly. The only opening is general legal education offering no services, and OpenAI reviews the landing page as well as the creative.

Does OpenAI allowing ads near legal and medical conversations mean those industries can advertise?

No. An April 2026 change stopped categorically blocking ads from appearing near medical, legal, and financial advice conversations. That governs placement, meaning which conversations can carry an ad from any advertiser. It does not govern permission, meaning who is eligible to buy. The two are frequently conflated.