Skip links
chatgpt ads for lawyers

ChatGPT Ads For Lawyers: Don’t Make This Mistake

ChatGPT Ads for Law Firms: Your State Bar Is the Real Gate

  • On August 31, 2026, OpenAI updated its ad policy to permit legal services ads in the US, but only when the advertiser is licensed to practice in the jurisdiction where the ad shows.
  • Every article about this covers OpenAI’s rulebook. Your state bar’s rulebook didn’t change, and it’s now the thing that decides whether you can run a campaign at all.
  • Florida makes you file each ad with the Bar 20 days before first use and pay $250, or $750 if you file late. That ends the “launch eight variants Monday, kill five Friday” workflow most PPC runs on.
  • Contextual targeting drops your ad under a conversation where someone is describing their accident in real time. No bar has ruled on whether that counts as advertising or solicitation.
  • Run a small, slow, documented test if you want the learning. Put the real money into being the firm ChatGPT names inside the answer.

For most of 2026, the answer to “can my firm advertise on ChatGPT” was no. OpenAI’s disallowed list put legal services next to gambling and political advocacy. Then on August 31 the policy flipped. OpenAI’s current ad policy permits ads for legal advice, representation, or legal services in the US when the advertiser is licensed to practice law in the jurisdiction where the ad is shown, and that language explicitly covers immigration, personal injury, legal claims, and document preparation. Legal services ads outside the US are still off.

Within about 72 hours, a dozen legal marketing agencies published some version of “the door is open, move now.” Fair enough. The part they skipped is that OpenAI was never the only gatekeeper. Your state bar is the other one, and nothing about its rules changed on August 31.

What actually changed, in plain terms

Legal services is a regulated category inside OpenAI’s system, which means manual case-by-case advertiser approval instead of instant onboarding, and you should expect to prove licensure during account verification. So “open” doesn’t mean you log in tonight and spend money tomorrow.

The scale is real. ChatGPT Ads hit a $1 billion annualized revenue run rate in under 200 days, with tens of thousands of advertisers on the platform. Self-serve access opened to US advertisers on May 5, and reported entry bids run around $3 to $5 per click with CPMs roughly in the $25 to $60 range. Against personal injury CPCs on Google, that looks like a steal.

Two limits before you get excited. Ads show only to logged-in adults on the Free and ChatGPT Go tiers. Plus, Pro, Business, Enterprise, and Education stay ad free. And ads still don’t appear in the sensitive conversations where a lot of your best cases actually start.

OpenAI also keeps moving the goalposts. The policy is on version 1.6 as of September 2026, and that update added language reserving OpenAI’s right to decline ads that conflict with its advertising principles, business interests, or competitive position. That’s the fourth or fifth revision since March. If you’ve watched Google rewrite the Local Service Ads rules mid-year, you know what building a channel on shifting policy feels like.

The bar rules bite harder here than on Google

This is the part that isn’t getting written about, and it’s the part that determines whether you can run this channel at all.

Filing requirements versus dynamic creative

Florida is the strictest example and I work with Florida firms, so start there. Most Florida lawyer ads have to be filed with the Bar’s Ethics and Advertising Department at least 20 days before first use, with a check for $250 per timely-filed advertisement and $750 for each one filed late. Firm websites are exempt. A sponsored placement inside ChatGPT is not a website.

Do the math on a normal test. You want four headlines against three descriptions. That’s twelve creative combinations, roughly $3,000 in filing fees, and three weeks of waiting before a single impression serves. Then your top performer loses to a variant you thought of in week two, and you file again.

Texas is gentler about timing. Under Texas Rule 7.04 you file a copy of the ad, an application, and the fee with the Advertising Review Committee within ten days of dissemination, with optional binding pre-approval if you submit 30 days ahead. Other states have no filing requirement at all. The rules genuinely differ, which is why the answer to “can I run this” is different in Tampa than in Tulsa.

The practical takeaway: whatever you build here has to be a small set of stable creatives you intend to run for months. The rapid-iteration habit from Google Ads doesn’t transfer.

The solicitation question nobody has answered

Google Ads match a keyword someone typed. ChatGPT ads match the conversation someone is having. Those aren’t the same thing, and the difference matters ethically.

Under Model Rule 7.3 and most state analogs, solicitation means a communication directed to a specific person the lawyer knows needs legal services in a particular matter. Now picture a user typing three paragraphs about getting rear-ended on I-95, their ER visit, and the adjuster who keeps calling. Your ad appears underneath. Did the platform “know” that person needs a lawyer for a particular matter?

I don’t know. Neither does your bar, because no state has issued an opinion on conversational AI ad targeting. Florida separately blocks written or recorded solicitation of a named accident or wrongful-death victim until more than 30 days have passed. Whether an automated contextual match falls inside that rule is untested.

Get a written opinion from your ethics counsel or your bar before you spend, not after someone files a grievance. That’s the whole recommendation. It costs a few hundred dollars and it’s the cheapest part of this entire channel.

Required disclosures in a very small box

The ad unit is tiny. A ChatGPT sponsored card carries a logo, a headline under 40 characters, a description block up to 150 characters, and a link.

Florida requires the ad to name a lawyer or the firm and the city of the principal office. New York requires most attorney ads to carry “Attorney Advertising” on the first page. Try fitting “Attorney Advertising” plus a firm name plus a city into 40 characters and see what’s left to sell with.

This isn’t fatal. It does mean your creative gets written by the compliance requirement first and the marketer second, which is the reverse of how most agencies draft ad copy.

Records you’re required to keep and can’t easily get

Florida also requires you to keep a copy of every advertisement for three years after you stop using it, along with records of when and where it ran.

Native reporting on ChatGPT ads is aggregate. You get impressions and clicks. You don’t get a served-creative archive with timestamps and placement detail. So build your own: screenshot every approved creative, log the launch and pause dates in a shared sheet, and export platform reporting monthly. Boring, and it’s the thing that saves you if a grievance lands in year two.

The licensing condition is also a targeting problem

Read OpenAI’s condition again. Licensed in the jurisdiction where the ad is shown.

That turns geographic targeting from an efficiency lever into a compliance control. If your attorneys are admitted in Florida only, a click from Georgia isn’t just wasted spend. It’s an ad for legal services shown somewhere you can’t practice.

Early reporting on the rollout described targeting as primarily contextual, with country-level geographic controls during the initial phase. More recent guidance suggests state-level targeting is available to legal advertisers. Confirm this inside Ads Manager yourself before you fund the account. If you can’t lock delivery to your admitted states, the channel isn’t usable for a single-state firm, no matter what the policy page says.

Ask your rep directly: what geographic granularity is available for legal services advertisers, and what happens to delivery in states outside my target?

What I’d do this quarter

A realistic plan for a firm that wants in without getting hurt:

  1. Get the ethics opinion first. Written, from your bar or your ethics counsel, on whether contextual AI ad placement is advertising or solicitation in your state.
  2. Apply for the advertiser account and start verification. Approval is manual and slow. Getting in line costs nothing.
  3. Write two creatives, not twenty. Compliant, stable, built to run for six months.
  4. File them if your state requires it, and budget the fees as part of the channel cost. In Florida, two ads is $500 before you’ve bought an impression.
  5. Confirm state-level geo controls in writing before funding.
  6. Build tracking yourself. Dedicated landing page, a unique tracking number, UTM tags. Platform reporting won’t tell you what signed.
  7. Cap the test. [JON: insert your standard test budget, e.g. $2,500 over 60 days] and a decision date.

Where the bigger opportunity still sits

Paid placement under an answer is one line of text. Being the firm ChatGPT names inside the answer costs nothing per click and doesn’t depend on a policy that’s changed five times this year.

That’s generative engine optimization, and it runs on the same inputs that have driven legal search for a decade: real reviews, citations that agree with each other, content that answers the actual question, and an entity the model recognizes. The firms winning organic AI recommendations today were doing that work in 2024.

Run the ad test for the learning. Keep the budget weighted toward the thing that compounds.

One more time, because it matters: none of this is legal advice, advertising rules vary a lot by state, and you should confirm everything above with your own bar before you spend.

If you want help deciding whether ChatGPT ads belong in your mix this quarter, or you’d rather put that money toward AI visibility instead, get in touch.

Related reading

This website uses cookies to improve your web experience.
Home