Last updated: July 21, 2026 · Checked against OpenAI's ad policy dated June 4, 2026

If you read our piece on whether personal injury firms can advertise on ChatGPT, you know the shape of the problem. OpenAI's policy names personal injury as prohibited, ads are running anyway, and the one open lane is general legal education that offers no legal services.

The obvious next thought, if you run mass tort, is that this is your lane. Everything that makes it a bad deal for a local firm looks like it reverses for you.

We thought so too. Then we read the rest of the policy.

Why mass tort looks like the exception

Start with the honest case, because it is a strong one.

Geography is not your problem. When ChatGPT ads launched, the worst thing about them for a local injury firm was that you could not target by state, so a Wisconsin firm paid for clicks from Arizona. OpenAI has since added targeting by state, DMA, and ZIP code, but that was never your concern either way. Mass tort is national by design. You are not licensed in one state trying to avoid the other forty-nine. The geographic limits that shaped a local firm's math simply do not apply to you.

You already run on education. The policy's carve-out permits general legal education and media where no legal services are offered. For a local firm, that means abandoning the case evaluation form and building content instead, which is a real change. Mass tort funnels already start with education. Do you have this condition. Did you take this drug. What does the litigation timeline look like. You are not being asked to do something foreign.

You already tolerate a long path. The education lane means the click lands on an article, not an intake form. For a firm used to measuring cost per signed case in a week, that is painful. Mass tort qualification already runs long, through screening, records, and criteria. A slower path is your normal.

Three for three. This is why the idea is worth taking seriously instead of dismissing. It is also why the actual answer is worth knowing before you spend anything.

A mass tort ad is two disallowed categories at once

Here is what the policy does to you.

The Disallowed Ads section prohibits legal services and names the examples: immigration, personal injury, legal claims, document preparation. Mass tort is legal claims. You are named directly, twice over, since mass tort is also personal injury.

That is the part everyone sees. Now the part nobody has written about.

OpenAI separately disallows health claims. Its ad content policy bars ads related to sensitive or regulated areas including health claims and healthcare, and it separately restricts claims related to the prevention, diagnosis, or treatment of a physical or mental health condition.

Read that against your creative. A mass tort ad is a claim that a drug or device caused a health condition. That is not a legal claim wearing a health costume. It is a causal assertion about a medical outcome, which is the definition of a health claim. Your ad says a product harmed people in a specific, diagnosable way.

And you cannot strip it out. The health claim is not decoration on a mass tort ad. It is the ad. Remove the assertion that the drug caused the injury and there is no campaign left.

So a local injury firm bumps into one prohibition. Mass tort bumps into two, and the second one has nothing to do with being a law firm. A pharmaceutical company making the same causal claim in reverse would hit the same wall.

The placement problem is worse than the content problem

Suppose you solve the creative. Suppose you build genuinely educational content, offer no representation, and get through review.

It still has nowhere to run.

Marketing Brew compared OpenAI's live ad policy against an April 29 version saved on the Wayback Machine and found something that has gone almost unnoticed. OpenAI added mental and personal health conversations to the list of sensitive contexts where ads cannot run. The earlier version did not treat personal health conversations as sensitive. It does now. An OpenAI spokesperson confirmed to Marketing Brew that the placement policy had been tightened, and reiterated that sensitive conversations remain barred from ads.

Now picture your prospect. They are asking ChatGPT whether their medication is linked to their diagnosis. Whether their symptoms match a known side effect. What a specific drug does to the kidneys.

Every one of those is a personal health conversation. That is not an edge case for mass tort. That is the entire funnel. The exact moment you want to reach someone is the moment OpenAI has decided no ad may appear.

The content policy blocks your creative. The placement policy blocks your context. Fixing the first does nothing about the second.

OpenAI has already told us what it considers safe

This is the part that should settle the question.

When the same OpenAI spokesperson described which contexts the company deems non-sensitive within regulated verticals, the examples given were general queries around exercise or diet, and recommendations for budgeting or financial education tools in response to broad financial questions.

Exercise and diet. Budgeting tools.

That is an OpenAI employee, on the record in a trade publication, drawing the line. Not "questions about drug side effects, handled carefully." Not "injury conversations with appropriate safeguards." Exercise and diet.

Measure the distance between that and someone describing their diagnosis to a chatbot. Anyone selling you mass tort placement inside ChatGPT is selling you the far side of a line OpenAI has already drawn in public.

The drug company cannot advertise either

Here is the detail that reframes the whole thing.

Pharmaceutical brands are not permitted in the ads pilot. Per the same Marketing Brew reporting, agencies representing pharma clients have been pushing OpenAI for access and have not gotten it. Those clients are instead advertising on physician-specific medical AI tools, because ChatGPT will not take them.

Sit with that. The company that made the drug cannot buy an ad next to a conversation about the drug. Neither can the firm suing them. Both sides of the most expensive advertising category in America are locked out of the same room.

So who appears in that conversation?

Whoever ChatGPT cites in the answer. That is the only party in the room, and there is no auction for the seat.

Why the education lane does not transfer

The lane works for a local injury firm because legal education can be genuinely legal. A guide to comparative negligence in Wisconsin is pure legal content. There is no health claim anywhere in it. It is dry, it is compliant, and it makes no assertion about anyone's medical condition.

Mass tort education cannot do that. Your educational content is about a drug and a diagnosis. That is health content by construction. The moment it is useful enough to attract the person you want, it is making the claim the policy prohibits.

And OpenAI reviews the creative and the landing page together. Its policy added a review scope section stating that ads are evaluated across creative and landing pages for compliance. So the trick of running clean creative and letting the landing page do the work does not survive contact with review.

The lane is open for legal education. It is not open for medical education. Mass tort only has the second kind.

What mass tort should actually do

For a local injury firm, our advice is that answer engine optimization beats the ad slot. For mass tort, it is stronger than that. AEO is not the better option. It is the only option, and it will stay that way longer than anyone expects.

Think about why. The ads are blocked by two separate policies, and the placement is blocked by a third. Even in the most optimistic scenario where OpenAI opens legal services tomorrow, the personal health conversation restriction sits untouched, and that restriction alone excludes your entire funnel. Legal opening does not help you. Health opening does not help you either, as long as the conversation itself is classified sensitive.

Meanwhile the conversations are happening right now, in volume, and the only thing that can appear inside them is a cited source.

That is not a consolation prize. It is an uncontested channel in the most contested vertical in legal advertising, and it is uncontested precisely because nobody can buy their way in. Not you, not the firm across the street, not the manufacturer.

The work is specific. Content that answers what people actually ask a chatbot about a drug and a diagnosis. Structured data that makes it machine-readable. Topical authority deep enough that a model reaches for you rather than a content mill. And measurement that connects a citation to a signed case, which almost nobody in legal marketing is doing.

The firms building that now will own those answers. The ones waiting for the ad platform to open are waiting for a door that, for mass tort specifically, may never open at all.

Frequently asked questions