Can Law Firms Use AI Generated Video in Ads? The Rule Is Not Symmetrical

Yes, a US law firm can run AI generated video ads, and the rule that governs it is not symmetrical. A synthetic person shown as a client is a disclosure problem you can solve. A synthetic person shown as the lawyer is closer to a flat prohibition, and no disclaimer is offered to cure it. Most vendors selling video to law firms never mention that distinction, and it is the one that decides whether a campaign is publishable.

That asymmetry is not an interpretation. It is written into California statute in two consecutive subdivisions, and the same result is reached in most other states through their rules of professional conduct. Here is what the text says, why the phrase "electronic image" matters more in 2026 than when it was drafted, and how to build an ad that stays clear of both edges.

The two subdivisions, side by side

California Business and Professions Code section 6157.2 lists things a lawyer advertisement may not contain. Two of them deal with portraying people who are not who they appear to be, and they are worded differently on purpose.

Subdivision (a)(3)(A) reaches "An impersonation of the name, voice, photograph, or electronic image of any person other than the lawyer, directly or implicitly purporting to be that of a lawyer." Read the end of that sentence and then look for the escape hatch. There is not one. The subdivision simply ends. Where a portrayal implies that the person on screen is a lawyer at your firm and they are not, there is no disclosure that makes it acceptable.

Subdivision (a)(3)(B) reaches "An impersonation of the name, voice, photograph, or electronic image of any person, directly or implicitly purporting to be a client of the licensee featured in the advertisement, or a dramatization of events, unless disclosure of the impersonation or dramatization is made in the advertisement." That final clause is the difference. Client portrayals and reenacted events are permitted, conditionally, and the condition is disclosure.

So the practical instruction for anyone generating video is short. Never let the presenter be read as your attorney. If the presenter is read as a client, or if the footage reenacts an event, say so on screen.

Why "electronic image" is the phrase that matters

That language predates generated video by decades. It was written for compositing, stock footage and lookalike casting. It is also broad enough to cover a person who never existed, because the statute is concerned with what the audience is led to believe, not with how the pixels were produced.

This is worth sitting with, because a common assumption inside firms runs the other way: that because the person is synthetic, no real person is impersonated, so the impersonation rules do not apply. The text does not support that. An impersonation is measured against the impression created in the viewer. A generated presenter in a suit, in front of law books, saying "we have recovered millions for our clients" implicitly purports to be a lawyer, whether or not any specific attorney was copied.

As of this writing no state has published a lawyer advertising rule that names generated video specifically. That is the honest position, and it cuts in an uncomfortable direction: the absence of a rule is not permission, it means a regulator will reach for the impersonation and dramatization language already on the books. If you want to see whether any authority in your own jurisdiction has addressed a portrayal question like this, you can search the case law in plain English rather than working through a citator by hand.

Three constructions that stay inside the line

Once the asymmetry is clear, most useful law firm creative is still available. These are the three shapes that do not depend on anyone's consent and can be produced on a schedule.

The first row is the one that scales. A narrator who explains what a contingency fee actually means, or how long a claim usually takes, is making claims the firm can substantiate and is not being held out as anyone. That is ordinary advertising, and it is the single largest body of useful content most firms never produce because filming a partner every week is not realistic.

Disclosure is a design problem, not a checkbox

Where disclosure is required, it has to be conspicuous, and that is where compliant intentions usually fail. Guidance across states treats a disclaimer as adequate only when it is clear and prominent, which in video practice means readable type held on screen for as long as the portrayal or claim is on screen. A single frame at the end does not do it. Small grey type behind a busy background does not do it.

Vertical formats make this worse in a way nobody anticipates. A 16:9 master gets trimmed to 9:16 for Reels, TikTok and Shorts, and the end card is exactly what gets cut. Firms routinely ship a compliant landscape ad and a non-compliant vertical version of the same thing. Build the disclosure into the template rather than adding it per asset, so that every export carries it by construction.

Two rules that catch firms out for unrelated reasons

While you are in the statute, two other subdivisions of section 6157.2 apply to every ad regardless of who is on camera.

Subdivision (a)(1) prohibits "A guarantee or warranty of success regarding the outcome of a legal matter as a result of representation by the licensee." The reach is wider than the word guarantee. A voiceover promising that the viewer will be taken care of, or a headline that asserts a result rather than an effort, can read as a guarantee to an ordinary viewer without the word appearing anywhere.

Subdivision (b) requires the advertisement to identify at least one lawyer licensed in the state, or the firm, responsible for the advertisement, together with office location information. That is a content requirement, not a formatting nicety, and it is the other thing that disappears when a landscape cut is cropped to vertical.

What this means for how you plan creative

The reason any of this matters commercially is that it decides what can be produced on a calendar and what cannot. Ads built on real client stories are excellent and they cannot be scheduled, because consent cannot be scheduled, and in most states you cannot ask a client to endorse you while their matter is pending. Ads built on what the firm knows can be produced every week, indefinitely, without a release from anyone.

That is the same split that runs through the economics of law firm advertising generally, where brand spend and intake spend answer to different clocks and different metrics. The intake half is bottlenecked by creative volume, and creative volume is the one constraint generated video actually removes.

Regulated verticals outside law hit the identical wall from a different direction. In healthcare the limiting factor is patient privacy rather than professional conduct rules, which is why the argument on the dental advertising page turns on written authorization instead of disclosure. The creative conclusion is the same in both: build the scalable half out of knowledge you own, and spend the consent-dependent half where trust is actually decided.

A short checklist before you publish

Run any generated law firm ad through five questions. Could a reasonable viewer think the presenter is a lawyer at this firm? Could they think the presenter is a real client? Does anything in the copy or voiceover promise an outcome? Is the required firm identification present on every aspect ratio you export? And if a dramatization or portrayal is used, does the disclosure stay on screen for its full duration in the vertical cut as well as the landscape one?

None of this is legal advice, and it is worth saying plainly that lawyer advertising is regulated state by state. Several states require advertisements to be filed or retained, some impose their own testimonial conditions, and two firms running identical creative in different states can get different answers. Read your own jurisdiction's advertising rule, and have counsel review anything that runs at scale.