Insights · AI search

What a ChatGPT ad can't buy your firm

ChatGPT now carries ads. For most advertisers, that is a new channel to test. For law firms, it is a narrow door, and it opens onto the wrong room.

Jacob Shamis
Jacob ShamisFounder and CEO · November 2026Last updated
A dark hallway with two openings: a narrow closet door, holding mops and hung with a brass sign reading Buy your ads here, and beside it a wide, brightly lit doorway into an open office

The paid door is narrow

OpenAI treats legal services as one of its most restricted ad categories. Its published ad policy, updated September 10, 2026, says ads for legal advice, representation, or legal services "are permitted in the US only when the advertiser is licensed to practice law in the jurisdiction where the ad is shown." It also lists legal services among the categories it is rolling out gradually, with each approval reviewed manually, case by case.

That is a reversal. Until late August the same policy barred legal-services ads outright, and trade press dates the change to August 31, 2026. Ads for legal services outside the US are still prohibited.

So the first question for a ChatGPT ad is not creative or budget. It is licensing. A ChatGPT ad is still a lawyer advertisement, under the advertising rules of every state where it runs. It is a state bar question before it is a marketing one.

And even when it clears, an ad is a placement. It is not a mention in the answer.

The old spend doesn't transfer

We have seen this before. In a June 2026 study across five AI engines, Morgan & Morgan, the country's largest personal-injury advertiser, was named in only a handful of AI answers (Citorian). Ad budgets did not carry over into the answer.

The answer is earned. The useful question is what earns it.

What AI actually cites

Otterly.AI, which tracks AI search, published a small and useful experiment on November 9, 2026. It took 60 topics and published each one four ways on LinkedIn: as a long-form article from a personal profile, as an article from a company page, and as a short post from each. It then tracked citations for 30 days across seven AI engines and 179 prompts.

Two results stand out.

  • Articles took almost everything. Of the 107 AI citations the campaign earned, 105 went to LinkedIn articles. That is 98.1%. Short posts earned two.
  • People beat the logo. Articles from the personal profile averaged 1.30 citations each. The same articles from the company page averaged 0.45. That is about 2.9 times more.

Read it with its limits, which Otterly states itself. It is one brand over one month. The personal profile had about 480 followers and the company page was new, so the 2.9x is not a clean test of person versus company. Nearly all the citations came from Microsoft Copilot and Perplexity.

The direction matches Otterly's larger study, published in June 2026, which covered about 1.3 million LinkedIn citations from January to June 2026. Articles drew 72.2% of citations to LinkedIn content, and 91.7% went to content from named individuals. Likes and comments showed no link to citations.

Why this matters more for law firms

AI answers about legal work name firms, practices, and people. A GC asks which firms handle a kind of matter, which practices do that work in a given city, and then checks the partner a referral produced. All three layers count. The lawyer layer is where firms are darkest, and it is where this data speaks most directly: the engines cite substantive writing under a person's name.

Your firm already has that expertise. Much of it sits in partners' heads, in client alerts, and in talks that never reached a page AI reads. A partner-authored article that answers a question your clients actually ask is the kind of source the engines cite. It is also a lawyer communication, so it goes through the same review as anything else your firm publishes.

This is a content strategy question, not a volume question. A few strong articles from the right lawyers, on the right questions, will likely do more than a steady stream of posts.

Measure, then decide

AI search moves fast. A separate tracker, cloro, found the share of AI answers citing LinkedIn fell by about 65% between February and July 2026. By October, short posts had overtaken articles among the LinkedIn pages cited. What works this quarter may not hold next quarter.

That is the case for measuring before you spend. Know what AI says about your firm, your practices, and your lawyers, on the questions your clients actually ask. See which sources the engines cite in your space, and whether your lawyers are among them. Then decide where the budget goes: a bar-cleared ad test, a partner article program, or fixing what AI already gets wrong about your firm.

Five things you can do now

  • Ask AI the questions your clients ask about your top practices, and note who gets named.
  • Fix the record AI cites: directory profiles, lawyer bios, your own practice pages.
  • Have two or three partners publish one long-form LinkedIn article each, from their own profile.
  • Clear any ChatGPT ad test with your ethics counsel before you spend.
  • Re-run the same questions every week and track what changes.

When the partnership asks what the firm is doing about AI, you will have the answer in your hand.


See what AI says about your firm

Run the free AI visibility check and see how the engines answer for your firm today.

See what AI says about your firm →

Sources

Related reading.