AI voice and avatar ads for medical practices

Generated voices and on-screen avatars are now cheap enough for a small practice to use. A vendor offers to clone the owner’s voice for outbound calls, to build a friendly video presenter in a white coat for social ads, or to read a script in a voice that sounds like a well-known actor. Each of these touches a different set of rules: the federal robocall law, the FTC’s rules on impersonation, endorsements and fake reviews, and state laws that protect a person’s voice and likeness.

This guide is for practice owners and health businesses deciding whether, and how, to use synthetic voices and people in advertising. It covers outbound calls, ads with avatars or synthetic clinicians, cloned voices of your own staff, and disclosure wording. AI-written posts and images in general are covered in our guide to AI-generated content in healthcare marketing. Disclosure laws for AI that talks with patients in chat, voice and letters are covered in our guide to state AI disclosure laws for patient messages. Real people hired to speak for the practice are covered in our guide to celebrity and spokesperson endorsements in practice ads.

This is general information, not legal advice. Sources cited were checked on October 30, 2026: the FCC’s February 2024 declaratory ruling on AI-generated voices (FCC 24-17), 47 CFR 64.1200, the FTC’s Rule on Impersonation of Government and Businesses (16 CFR Part 461), the FTC Endorsement Guides (16 CFR Part 255), the FTC rule on consumer reviews and testimonials (16 CFR Part 465), Tennessee Public Chapter 588 of 2024, California Civil Code 3344, New York Civil Rights Law 50 and New York General Business Law 396-b as amended by S.8420-A. State rules on voice, likeness and AI disclosure differ, so check yours, and have a health care attorney review a synthetic voice or avatar campaign before it runs.

The short version

Outbound calls: an AI-generated voice is an “artificial” voice under the Telephone Consumer Protection Act, so the consent rules for prerecorded calls apply. Promotional calls need prior express written consent.
Synthetic clinicians: an avatar presented as your physician, nurse or another expert must reflect a real person with those qualifications, who agreed to the use and holds the views stated.
Synthetic patients: an AI-made patient presented as a real patient is a fake testimonial. Use real patients with consent, or say clearly that the people shown are not patients.
Real voices and faces: using a real person’s voice or likeness to advertise without consent can violate state publicity and privacy laws, including laws that now name simulated voices.
Disclosure: New York now requires a conspicuous disclosure when an ad includes a synthetic performer. Elsewhere, a clear disclosure is still the plain way to avoid a misleading net impression.

Outbound calls with an AI voice

On February 2, 2024, the FCC adopted a declaratory ruling, released February 8, 2024 as FCC 24-17, that answers the first question practices ask. The Commission confirmed that the TCPA’s restrictions on the use of an “artificial or prerecorded voice” cover current AI technologies that resemble human voices or generate call content using a prerecorded voice. It named voice cloning as one of them, because a person is not speaking the message. The ruling took effect on release.

The ruling also closed a route a vendor may pitch. Relying on its earlier soundboard ruling, the FCC said the TCPA does not allow a carve out for technologies that claim to be the equivalent of a live agent, and that this applies whether the AI voice is wholly synthetic or resembles a real person’s voice taken from an audio clip.

What that means under 47 CFR 64.1200:

Promotional calls need written consent. A call to a residential line or a cell phone that uses an artificial or prerecorded voice and includes or introduces an advertisement or constitutes telemarketing needs the called party’s prior express written consent, under 64.1200(a)(2) and (a)(3). The rule defines that as a signed written agreement, which can be electronic, that clearly authorizes the seller to deliver such calls to a specific number, with the disclosures the rule lists.
The caller must identify itself. Under 64.1200(b)(1) and (b)(2), each artificial or prerecorded voice message must state clearly at the beginning the identity of the business responsible for the call, using its registered name, and must state a callback telephone number during or after the message.
Ad calls need an automated opt-out. Under 64.1200(b)(3), a message that includes or introduces an advertisement or constitutes telemarketing must offer an automated, voice or key press opt-out during the call, and a toll free number to reach it when the message is left on voicemail.

The rule contains a narrow allowance for calls that deliver a “health care” message made by or on behalf of a HIPAA covered entity or its business associate, with its own conditions. Whether a given call fits that allowance is a question for your attorney. A call that promotes a new service, a seasonal special or a membership plan is the kind of call to treat as an ad that needs written consent. Our guide to business texting rules for medical practices covers the parallel rules for texts.

Use: “This is an automated call from [Practice name], at [phone]. You asked us to call about [topic]. Press 9 to stop these calls.” Avoid: a cloned voice that opens with “Hi, it’s Dr. [Name]” with no word that the voice is recorded or generated, placed to a list with no written consent on file.

Synthetic clinicians and the impersonation rule

The FTC’s Rule on Impersonation of Government and Businesses, 16 CFR Part 461, took effect in 2024 (published at 89 FR 15030 on March 1, 2024). Section 461.3 makes it an unfair or deceptive practice to materially and falsely pose as a business or an officer of a business, or to materially misrepresent affiliation with, endorsement by or sponsorship by a business or its officer. Section 461.2 says the same about government entities and officials. The rule defines “officer” to include executives, officials, employees and agents, and defines “business” to include not-for-profit entities, which covers hospitals and health systems.

Two practical limits on scope. First, as of the eCFR text current in October 2026, Part 461 contains sections on government and business impersonation only. It does not contain a general ban on impersonating private individuals. Second, the rule turns on material misrepresentation, so the question is what a viewer would reasonably take away.

For a practice, the rule matters when an ad uses a generated voice or face to suggest a connection that does not exist:

A synthetic presenter introduced as “a physician at [Hospital name]” when no such physician endorsed the service.
An avatar in a uniform or with a seal that suggests a government health agency backs your clinic.
A cloned voice of a well-known health system’s spokesperson promoting your practice.

Even outside Part 461, Section 5 of the FTC Act, 15 U.S.C. 45, prohibits deceptive acts or practices in commerce. A synthetic person who misleads viewers about who is speaking can be a deceptive practice on its own.

Synthetic doctors and the endorsement guides

The FTC Endorsement Guides, 16 CFR Part 255, define an endorsement in 255.0(b) as an advertising message that consumers are likely to believe reflects the opinions, beliefs, findings or experiences of a party other than the sponsoring advertiser. Depictions of the name, likeness or other identifying personal characteristics of an individual can be endorsements, and the endorser “could be or appear to be” an individual, group or institution. A generated person in a white coat who praises your treatment fits that definition.

The guides then set requirements that a synthetic presenter cannot meet on its own:

Honest opinion and real use. 255.1(a) says endorsements must reflect the honest opinions, findings, beliefs or experience of the endorser. 255.1(c) says that when an ad represents that the endorser used the product, the endorser must have been a bona fide user. A generated character has no opinions and no experience.
Real expertise. 255.3(a) says that when an ad represents the endorser as an expert, the endorser’s qualifications must in fact give that expertise, and 255.3(b) requires an actual exercise of that expertise in evaluating the product. An avatar presented as a dermatologist or a dentist is represented as an expert it is not.
Advertiser liability. 255.1(d) says advertisers are subject to liability for misleading or unsubstantiated statements made through endorsements.

This does not rule out a presenter. 255.0 includes an example of an announcer who does not claim to speak from personal experience and speaks on behalf of the company, and says that is not an endorsement. A generated narrator who plainly speaks for [Practice name], without claiming to be a clinician, a patient or a user, is closer to that announcer. The clinical claims in the script still need the same support as any other ad claim, which our guide to healthcare marketing rules covers.

Using your own clinician’s likeness is different. A synthetic version of Dr. [Name], made with that physician’s written consent, saying only what the physician has reviewed and holds as their view, is still the physician’s endorsement. Keep the approved script, the consent and the date on file, and stop running it if the physician leaves or changes their view, because 255.1(b) asks advertisers to have good reason to believe the endorser still holds the views presented.

Synthetic patients and fake testimonials

A generated patient telling a recovery story is the clearest problem. Two federal rules apply.

16 CFR 255.2(c). Ads presenting endorsements by what are represented, expressly or by implication, to be actual consumers should use actual consumers in both the audio and the video, or clearly and conspicuously disclose that the people in the ads are not actual consumers of the advertised product.
16 CFR 465.2. The FTC’s rule on consumer reviews and testimonials makes it an unfair or deceptive practice for a business to write, create or sell a consumer testimonial that materially misrepresents that the testimonialist exists, that they used the service, or what their experience was. It also bars disseminating a testimonial the business knew or should have known was misrepresented in those ways.

A disclosure that the person is not a real patient answers the actual consumer question under 255.2(c). It does not make an invented recovery story an acceptable claim about results, because 255.2(a) and (b) treat consumer endorsements about performance as claims the advertiser must substantiate, including whether the experience shown is representative. For health outcomes, that means scientific support, not a script. Real patient stories, consent and before and after rules are covered in our guide to patient testimonials and before and after photos.

Use: “Actor portrayal. Not a patient.” with a general description of the visit, such as what the first appointment looks like. Avoid: a generated “patient” saying “I lost 40 pounds in 3 months here”, with or without a disclosure.

State laws on voice and likeness

State laws protect a person’s name, voice and likeness from commercial use without consent, and some now address simulated voices directly. Three examples, each checked against its official text:

Tennessee. Public Chapter 588 of 2024, the Ensuring Likeness, Voice, and Image Security Act, effective July 1, 2024, amended Tenn. Code 47-25-1101 and following. It defines “voice” as a sound readily identifiable and attributable to a particular individual, whether it contains the actual voice or a simulation of it, gives each individual a property right in their name, photograph, voice and likeness, and creates civil liability for using them in advertising without prior consent. It also reaches a person who makes available a tool whose primary purpose is producing a particular person’s voice or likeness, knowing the use was not authorized.
California. Civil Code 3344 makes a person who knowingly uses another’s name, voice, signature, photograph or likeness for advertising or selling, without prior consent, liable for damages of the greater of 750 dollars or actual damages, plus profits from the use, attorney’s fees and costs, and possibly punitive damages.
New York. Civil Rights Law 50 makes it a misdemeanor to use the name, portrait, picture, likeness or voice of a living person for advertising or trade without written consent. Separately, General Business Law 396-b, as amended by S.8420-A, requires anyone who produces or creates an ad for a commercial purpose to disclose conspicuously that a synthetic performer is in the ad, where they have actual knowledge. The bill text defines a synthetic performer as a digitally created asset, made with generative AI or a software algorithm, intended to look like a human performer who is not recognizable as any identifiable real person. It sets civil penalties of 1,000 dollars for a first violation and 5,000 dollars for each later one, excludes audio ads and AI used only to translate a human performer, and took effect 180 days after becoming law. The bill was signed in December 2025, which places the start date in June 2026; confirm the current text with your attorney.

Two points follow. A voice that sounds like a well-known actor or a local news anchor is a risk even if no recording of that person was used, because Tennessee’s definition reaches a simulation of an identifiable voice. And your own staff members have these rights too. Cloning a front desk lead’s voice for reminder calls needs that person’s written consent, a clear scope and a plan for when they leave.

Other states have their own rules. State rules on voice and likeness differ, so check yours before a campaign runs where your patients live.

Disclosure wording that works

Where a disclosure is required or prudent, the FTC’s standard in 16 CFR 465.1(c) gives a usable test: a clear and conspicuous disclosure is easily noticed and easily understood. In a visual ad it goes on screen long enough to read; in an ad with both audio and visuals it appears at least in the same means as the claim it qualifies; in audio it is spoken at a volume, speed and cadence ordinary listeners can follow.

Use:

“This presenter is AI-generated. Not a real person.” on screen for the whole time the presenter speaks.
“Voice generated with AI on behalf of [Practice name].” at the start of a call or audio spot.
“AI-generated likeness of Dr. [Name], used with Dr. [Name]’s permission. Script reviewed by Dr. [Name].”
“Actor portrayal. Not a patient.”

Avoid:

A disclosure in small gray text in the closing frame, or only in the ad’s caption.
“Enhanced with technology” or similar wording that does not tell the viewer the person is not real.
Calling an avatar “our doctor” or “our nurse” when no such clinician stands behind it.
A disclosure only in a linked page when the ad itself makes the claim.

Ad platforms may set their own rules for health ads and for altered or synthetic media, and those rules change. Check the current policy of each platform you use before a campaign launches.

AI front desk voices are a separate question

An AI receptionist that answers inbound calls is not an outbound robocall, and it is not an ad. Its rules are different: what it may say about itself under state AI disclosure laws, how it hands calls to staff, and how it handles patient information. Those are covered in our guides to AI receptionists for medical offices and state AI disclosure laws. If the same voice later places outbound calls that promote services, the outbound rules in this guide apply to those calls.

An illustration

This example is invented. [Practice name], a med spa, wants to launch a new skin treatment. A vendor proposes three things: a 30 second video with an AI presenter in scrubs introduced as “our aesthetic nurse”, two AI “clients” describing results, and outbound calls in a cloned version of the owner’s voice to the full patient list.

Their attorney sends it back. No nurse stands behind the presenter, the “clients” do not exist, and the call list has no written consent for promotional calls.

The revised plan: the real nurse injector records a short video in their own voice about what a first visit involves, with no promised results. The AI presenter stays only as a narrator for an explainer on aftercare, labeled “AI-generated presenter” on screen. The testimonial clips are dropped until real clients agree in writing. Outbound calls go only to patients who signed a written consent for promotional calls, open with the practice name and number, and offer a key press opt-out. The owner signs a written consent covering use of their voice for those calls only.

Checklist before a synthetic voice or avatar runs

Each outbound call with an AI or prerecorded voice that promotes services has prior express written consent on file for that number.
Each call message names the practice at the start, gives a callback number, and offers the automated opt-out for ad calls.
No synthetic person is presented as a clinician, patient or expert unless a real person with those qualifications stands behind it with written consent.
No synthetic person suggests a link to a hospital, health system, agency or brand that has not agreed to it.
No real person’s voice or likeness, including staff and sound-alikes of known people, is used without written consent.
Each clinical claim in the script has support, and patient outcomes shown are real and representative.
A clear disclosure appears in the same medium as the synthetic content, and the New York disclosure is in place for ads shown there.
Consents, scripts, review dates and disclosures are filed with the campaign.

How we help

We build websites, local search, Google Ads and lead generation and AI front desk tools for independent practices and health businesses. For ads, that means campaigns built on real clinicians and real patient stories with consent, consent capture forms that record what a patient agreed to and when, and clear labels wherever a generated voice or image is used. Our tools are built with HIPAA in mind. We confirm in writing what a BAA covers before any patient information flows through a tool, and if it is not covered, we do not route patient information through it. We are not a healthcare provider and we do not give legal or medical advice. For the wider rules, see our healthcare marketing rules hub.

Can a medical practice use an AI voice for marketing calls?

Only with the right consent. In FCC 24-17 (February 2024) the FCC confirmed that AI-generated voices are artificial voices under the TCPA. Under 47 CFR 64.1200, calls with an artificial or prerecorded voice that advertise or telemarket need the called party’s prior express written consent, must identify the caller at the start and must offer an automated opt-out.

Can we use an AI avatar as a doctor in our ads?

Not as a doctor who does not exist. The FTC Endorsement Guides require expert endorsers to have the expertise they are presented as having and to actually exercise it. A generated narrator who speaks for the practice without claiming to be a clinician, clearly labeled, is a different matter.

Can we clone our own physician’s voice or face?

With the physician’s written consent, for a defined use, saying only what they have reviewed and hold as their view. State laws such as California Civil Code 3344, New York Civil Rights Law 50 and Tennessee’s 2024 voice law protect a person’s voice and likeness from commercial use without consent.

Is an AI-generated patient testimonial allowed if we label it?

A label can address whether the person is a real patient, under 16 CFR 255.2(c), but the outcome described still has to be real, supported and representative. 16 CFR 465.2 bars creating testimonials that misrepresent that the testimonialist exists or what their experience was.

Does New York require a disclosure for AI people in ads?

Yes. General Business Law 396-b, as amended by S.8420-A, requires a conspicuous disclosure when an ad includes a synthetic performer and the advertiser knows it, with civil penalties of 1,000 dollars for a first violation and 5,000 dollars after that. Audio ads are excluded.

AI voices, avatars and synthetic clinicians in practice ads: the rules that apply