APTA names AI 3 times. Not once with teeth.

Three. That's how many separate places physical therapy's brand-new Code of Ethics names artificial intelligence by name, more than the occupational therapy or speech-language pathology codes covering the same ground. If you've been putting off using AI to draft your clinic's Google Business posts, Instagram captions, or review responses because you assumed a rule against it was buried somewhere in APTA's ethics code, there mostly isn't one. The rule that actually has a number attached to it comes from a federal agency that has never treated a patient.

1. Does APTA's Code of Ethics say anything about using AI in marketing?

Three things, and none of them are what a practice owner worrying about liability actually needs. APTA's Code of Ethics for the Physical Therapy Profession, effective January 1, 2026 (APTA, direct PDF fetch, checked September 2026), names AI in Standard 3.D (accountability "for the accuracy and truthfulness of information they disseminate, including in the use of emerging technologies, such as social media and artificial intelligence"), Standard 5.C (maintaining professional responsibility "when information is disseminated using current and emerging technologies"), and Standard 8.A (developing competence in current technologies "including but not limited to social media and artificial intelligence"). All three sit in the Code's aspirational tier, illustrative examples to strive toward, not enforceable Standards of Conduct the APTA Ethics and Judicial Committee disciplines members under. The one enforceable marketing rule, Standard 6.1, requires that PTs "provide information about their services that is truthful and accurate" and make no misleading representations "in any forms of communication, including billing." It does not mention AI, because it does not need to.

So what for you: the standard you can actually be disciplined under already covered AI-drafted marketing before AI existed, because it regulates the claim, not the tool that typed it. The AI-specific language exists only in the aspirational tier, which is guidance to strive for, not a rule with a complaint process behind it.

2. Can I use AI to write or polish a patient testimonial?

Not without the patient reviewing and signing off on the exact final wording, and the precedent predates AI by a decade. In 2016, the HHS Office for Civil Rights settled with a Los Angeles physical therapy practice, Complete P.T., Pool & Land Physical Therapy, for $25,000 after it posted patient testimonials online, complete with full names and photos, without valid HIPAA authorization (HHS Office for Civil Rights resolution agreement, announced February 16, 2016, following a complaint filed August 2012). The practice admitted civil liability and ran a three-year corrective action plan. Nothing about that exposure changed with AI, but an AI tool that "polishes" a two-line patient comment into a glowing testimonial, or drafts one from scratch and attributes it to a real name, can cross the same line at higher volume than a front-desk staffer typing it out by hand. The FTC's Consumer Reviews and Testimonials Rule (16 CFR Part 465) adds a second exposure point, prohibiting AI-generated endorsements that misrepresent what a reviewer actually experienced.

So what for you: never let an AI tool generate or embellish testimonial language attributed to a named patient before that patient has reviewed and approved the exact wording in writing. That's the same authorization step a California practice got fined for skipping in 2012, now easier to skip by accident.

3. What's the actual penalty for posting AI-generated fake reviews?

Up to $53,088 per violation, and this is the number that should actually shape a marketing policy. The FTC's Trade Regulation Rule on Consumer Reviews and Testimonials (16 CFR Part 465, effective October 2024) prohibits AI-generated fake or deceptive reviews and testimonials outright. Knowing violations of that rule carry a civil penalty capped at $53,088 per violation, set by the FTC's January 2025 inflation adjustment and confirmed unchanged for 2026 by the Federal Register (7 July 2026). That figure attaches to the specific act of an AI-generated fake review or testimonial, not to AI-drafted marketing copy in general, but it is assessed per violation, not per campaign, and it applies whether or not a practice belongs to APTA at all. No association ethics code carries a comparable civil penalty; a Standard 6.1 violation goes through APTA's own disciplinary process, not a federal fine.

So what for you: treat the FTC's rule, not your association's code, as the one with real teeth, and build your AI content policy around avoiding a $53,088 exposure rather than an ethics complaint that carries no equivalent dollar figure.

4. Do I have to disclose that a post was written by AI?

For a review or testimonial, yes. For a general blog post or service description, not automatically, but disclosing costs nothing and removes the question entirely. The FTC's Endorsement Guides require a clear and conspicuous disclosure whenever a connection between an endorser and a business could affect how an audience weighs the endorsement, a standard the FTC confirmed in 2024 covers AI-generated content as well as human endorsers. Routine educational content, a post explaining what a first evaluation involves or how to manage post-op swelling, doesn't carry PHI or make a personal endorsement, so it falls outside that specific disclosure trigger. APTA's aspirational Standards 3.D and 5.C point the same direction without creating a separate enforceable duty: accuracy and accountability when AI is involved in disseminating information.

So what for you: build a one-line AI disclosure into any content template that resembles a review, testimonial, or endorsement now, before volume makes it easy to forget on the one post where it actually mattered.

5. What does an AI marketing stack actually cost for a small practice?

A fraction of what a marketing vendor's first quote usually suggests. Physical therapy marketing agencies that publish their own pricing, serving one to three clinic practices, run $199 to $3,095 a month plus $0 to $2,500 in setup fees, with SEO packages starting around $300 a month on their own (composite of published PT marketing agency pricing pages, checked September 2026, vendor-sourced, worth confirming directly before committing). A solo or small-practice AI marketing stack looks closer to Canva Pro at $15 a month for design, Buffer from $5 per connected channel a month for scheduling, with its AI assistant included free on every plan, and ChatGPT Plus at $20 a month for drafting, roughly $30 to $50 a month total (published vendor pricing, September 2026, treat as a snapshot).

So what for you: if the first number a marketing vendor quotes runs into four figures a month, you're likely looking at a package built for a multi-site group, not what a three-clinician independent practice needs to get started.

6. What is answer engine optimization and should I care about it yet?

It's a genuinely different skill from traditional search engine optimization, and it rewards structure over ad spend. Traditional SEO competes for a clicked position on a results page. Answer engine optimization competes to be the source an AI platform, ChatGPT, Perplexity, Google's AI Overviews, actually cites inside the answer it generates before the searcher clicks anything. The levers that consistently show up across 2026 marketing guidance on this (composite of industry AEO guides, checked September 2026, informed practice rather than a controlled study): a short, direct answer near the top of each key page, 40 to 60 words, FAQ content marked up so a model can parse question and answer pairs directly, and specific named details, credentials, services offered, location, rather than generic marketing language.

So what for you: before paying for a dedicated AEO tool, check whether your own service pages already answer a direct question in the first few sentences. That's a free structural fix, and it's the exact thing these guides are charging to do for you.

The bottom line

Two rules matter here, and they're not the ones most practice owners are watching. Standard 6.1 already required truthful, non-misleading marketing before AI existed, and it still does regardless of what tool drafted the sentence. The FTC's Reviews and Testimonials Rule carries the actual dollar figure, up to $53,088 per violation, and it doesn't care whether you've ever paid APTA a membership fee. Build a policy around three things: keep content accurate under Standard 6.1, get written patient sign-off before any testimonial goes out, and add a disclosure line to anything resembling a review. That covers both exposure points for less than a single marketing agency setup fee. For the same enforceable-versus-aspirational gap in occupational therapy and speech-language pathology, see $53,088 a post: FTC's rule, not AOTA's. For the enterprise-versus-affordable pricing pattern in another specialty, see $39, not $249: the dental AI marketing myth. If marketing turns out to be one gap among several, the AI Opportunity & Growth Assessment scores it against the CARE Framework.

Related: $53,088 a post: FTC's rule, not AOTA's  ·  $39, not $249: the dental AI marketing myth  ·  $670-$1,345: AI audit for physio practices

This article is for informational purposes only and does not constitute legal advice. Federal, state, and association rules change frequently; confirm current requirements with legal counsel and your professional association before relying on anything above.