AI Lawyer Advertising Rules: Bios, Results Claims, Chatbots and Rule 7.1

Suppose an associate asks a chatbot to “polish” a partner’s website bio and the draft comes back with a Chambers ranking the partner never held. Nobody typed it in. Nobody notices. It goes live. Under ABA Model Rule 7.1 that is a false communication about a lawyer’s services, and who wrote it never arises. The best one-line summary of AI lawyer advertising rules comes from an overview by the marketing firm ThriveSearch: “The tool is new; the obligation is not.” And: “The ‘AI drafted it’ defense does not exist.”

It is not far-fetched. In Concord Music v. Anthropic a Latham & Watkins associate asked Claude “to provide a properly formatted legal citation” for a real article; the link was right, the author and title were wrong, and the “manual citation check did not catch that error”. A formatting request still generates content. Ask a model to tidy a bio and it may tidy in an award.

The rule: Rule 7.1 does not care who wrote the sentence

Rule 7.1 prohibits false or misleading communications about the lawyer or the lawyer’s services; it looks at the words, not the author. Micah Buchdahl, writing for ABA Law Practice, lists the rules AI marketing touches (1.1, 1.6, 5.3, 5.4, 5.5, 7.1 and 7.3) and sums up the problem: “AI just makes it up.”

Three fabrications account for most of the risk: credentials, case results and testimonials, such as “inventing a ‘top-rated’ award or recognition the firm never received” (ThriveSearch). LaFleur’s 50-state overview reads the New York State Bar’s 2024 task force report as treating AI content as attorney speech and ruling out synthetic endorsements. In England the SRA’s warning notice of 17 August 2026 says solicitors “remain accountable for their work and outputs, regardless of how that work has been prepared”. Nothing in it exempts a website.

Bios: hallucinated credentials, and five prompts that do not invent any

Keith Wewe of Content Pilot published five ChatGPT prompts for lawyer bios that work because none of them asks the model to add facts. One: “Do the first 2-3 sentences describe the lawyer’s practice and the type, size, location, and industry sector of clients represented?” The others test active voice, conversational tone, scannability and client focus. Combined, with a verification table at the end:

Audit the bio below against five tests and rewrite where it fails: (1) do the first two or three sentences state the practice and the type, size, location and industry of clients served; (2) client-focused rather than a CV; (3) conversational, active voice; (4) scannable, with a representative-matters list; (5) would an AI assistant asked "who handles [practice] in [city]" find the specialisation and location signals.
Rules: use only facts in the bio. Do not add, upgrade or generalise any credential, award, ranking, result or client name. Where a fact is missing, write [NEEDS FACT].
Output: the rewrite, then a table of every credential, award, result and client reference you kept, each marked "verify against: [source]".

Bio:
[paste]

Results claims, testimonials and “our AI is better than theirs”

Two things go wrong. Models quietly upgrade claims: the tell-tale edit is “leading” becoming “the leading”, which is why any AI rewrite of marketing copy needs a change log of every sentence whose meaning shifted. And models manufacture social proof on request: ask for “three client testimonials” and you will get three. A testimonial no client gave is a false statement under Rule 7.1, whoever typed it.

Florida sets the clearest bar. Ethics Opinion 24-1 (19 January 2024) says lawyers “cannot claim their generative AI is superior to those used by other lawyers or law firms unless the lawyer’s claims are objectively verifiable”. The temptation is a line saying your AI makes you faster than other firms; you would need a benchmark, not a vendor brochure. Say what you do and how you check it, and leave the adjectives out.

Review the marketing copy below against [state / SRA] lawyer-advertising rules and Rule 7.1. List every factual claim, comparative or superlative ("leading", "best", "top-rated"), result ("won", "recovered", "secured"), testimonial, credential and guarantee. For each: the substantiation we would need on file; whether a qualifier or date is needed; and a compliant rewrite that keeps the marketing voice.
Flag separately: any named client; any implied outcome; any claim that our AI tools are better or faster than other firms'.
Do not add claims. Do not soften more than necessary.

Copy:
[paste]

Chatbots on your website: the disclosure rules

Intake chatbots must disclose that they are AI in New York, California, Colorado, Utah and New Jersey, according to ThriveSearch’s overview; New Jersey’s test is whether “a reasonable person might be misled about whether a human is participating”. Florida 24-1 tells lawyers to be “wary of utilising an overly welcoming generative AI chatbot that may provide legal advice, fail to immediately identify itself as a chatbot, or fail to include clear and reasonably understandable disclaimers”. In the EU, the AI Act’s Article 50 transparency obligations apply from 2 August 2026; the Digital Omnibus of July 2026 left that date alone and deferred only the Article 50(2) watermarking duty for systems already on the market, to 2 December 2026. BRAK’s guidance adds that intake chatbots carry a heightened duty of care.

ThriveSearch’s sample opening line does the job: “I’m an AI assistant, not a lawyer. I can share general information about this firm’s services and help you schedule a consultation. I don’t provide legal advice, and chatting here doesn’t create an attorney-client relationship.”

You are the intake assistant for [firm], a [practice areas] firm in [city, state]. Open every conversation with the AI disclosure in <disclosure>.
You may: describe practice areas and offices from <firm_facts>; collect name, contact details, matter type and urgency; offer consultation slots from <calendar>.
You may not: assess the merits of anyone's case, estimate outcomes or values, quote fees beyond <fee_page>, run a conflict check, or state any award, ranking or result not in <firm_facts>.
If asked for legal advice, say a lawyer will call. If someone mentions an emergency or a deadline, mark the handover note URGENT.

Florida 24-1 and the other rules, side by side

By one 2026 count, more than twenty-five state bars have AI guidance; the state-by-state map covers them and the Article 4 guide covers the EU. Whether you must tell clients you use AI at all is answered in the disclosure guide.

Images, video and synthetic people

New York requires disclosure where synthetic performers appear in commercial advertising, per ThriveSearch; the EU’s Article 50 covers AI-generated image, audio and video from August 2026. Copyright is calmer than the headlines: in Getty Images v Stability AI [2025] EWHC 2863 (Ch) the English High Court rejected Getty’s secondary-infringement claim because the model does not store or reproduce its training works. Rule 7.1 survives: a generated image of a beaming “client” outside a courthouse implies a result and a person that do not exist.

Why the brand is worth this much care

Originality.AI classified 81.2% of 5,000 July 2026 LinkedIn posts as “Likely AI”, and LinkedIn is rolling out a “Seems like AI slop” feedback option. Meanwhile Passle’s Tom Elgar reports direct traffic to legal thought leadership up 59% while The Economist found search visits to reference sites down 15%: “AI thrives on yesterday’s internet. Lawyers who are winning the content game write about tomorrow’s legal risk.” Credibility is the asset the rules protect; the LinkedIn guide and the GEO guide show how to keep it.

The “Roger” story fits here. A trial lawyer prepared for twelve GC meetings in two hours with ChatGPT, Lexis+ and Introhive and, per a single vendor account from the Legal Sales and Service Organization, won new opportunities from eleven. The AI did the research; Roger did the talking. That division of labour is AI marketing ethics in one sentence.

The pre-publication checklist

Axiom’s Andra Dallas, on reviewing AI marketing content: “Don’t just review things by email. Don’t just review things by Slack. There has to be a system in place for review.” Ten lines:

1. Every credential, award, ranking and result traced to a source document on file.
2. Every testimonial matches a real client’s written words and consent.
3. No superlative or comparative claim without objective support.
4. No named client without permission; no implied outcome in copy or images.
5. The chatbot opens with an AI disclosure and a no-advice statement everywhere you serve.
6. Synthetic images, voices or performers labelled where required (New York; EU Article 50).
7. Jurisdiction-specific rules checked for every state or bar the campaign targets.
8. Press releases and crisis statements approved in writing by the matter partner, practice-group leader or general counsel, as Jaffe PR’s policy template requires.
9. The named lawyer has read the final text and can defend every sentence.
10. Log who reviewed it and when; if you label anything, “AI-assisted and attorney-reviewed” is the label.

Or, as rankings.io puts it, “AI is a drafting tool, not a compliance shortcut.” The business-development module of AI Lab for Lawyers covers LinkedIn and outreach drafting hands-on; this checklist is the review step to apply to whatever you build there. The three prompts above sit in the prompt library.

Where to go next: the business-development pillar for the full BD workflow, the RFP guide for the pitch-side fabrication trap, and client communication for the engagement-letter side. The rest of the cluster is in the business-development hub.