loading workspace…
A model paraphrase you didn't write can still trip Rule 7.1. A practical framework for substantiation logs and partner sign-off when AI re-renders your claims.
ABA Model Rule 7.1 prohibits false or misleading communications about a lawyer or the lawyer's services. The state versions track closely. For decades, the operative question has been: what did the firm publish, and is it substantiated?
That question has a new wrinkle. When ChatGPT paraphrases your firm's practice description — adding nuance, dropping caveats, or inferring claims you never made — and a prospective client relies on that paraphrase, is the firm responsible for what the model said? Most of the time, yes. Almost no firm is set up to handle that risk.
Rule 7.1's text governs "a communication about the lawyer or the lawyer's services." Comment [1] makes clear that communications include anything material that creates an unjustified expectation. The relevant question is not whether the lawyer typed the words, but whether the communication is attributable to the lawyer.
When a model paraphrases your firm's own published content, the paraphrase derives from your communication. State bar advisory committees have started taking the view — informally so far — that the originating firm is responsible for foreseeable paraphrases of its published content. Foreseeability is the operative word.
No issue. The model summarizes your "30+ patent litigations before the ITC" claim into "extensive ITC patent litigation experience." Same meaning, no inflation. You're fine.
The model takes your "30+ patent litigations" and renders it as "leading patent litigation boutique." You did not claim market leadership; the model inferred it. This is where the risk lives. Under the foreseeability standard, if a reasonable firm could anticipate that an LLM would re-render this kind of claim, and the firm did not take steps to ground the model's re-render, you may be responsible.
The defensive posture: substantiation logs and explicit non-comparative language in the source content.
Your site copy says "trusted by Fortune 500 clients." The model renders that as "leading Fortune 500 IP counsel." You started in 7.1-borderline territory; the model pushed it into 7.1-violation territory. Joint exposure: your borderline copy is the originating issue.
The model hallucinates. It tells a prospect you have "twelve partners admitted before the Supreme Court" when you have one. This is the only scenario where the lawyer is squarely not responsible — the model is operating outside the foreseeable space of paraphrase. But you still have a remediation obligation: notify the prospect, document the correction, update the source content if there is ambiguity that invited the hallucination.
Every substantive marketing claim on a firm site should have:
The last item is novel and is the one most firms miss. If your claim is "30+ patent litigations before the ITC," your paraphrase envelope might be:
At Cognoverge we run every recommendation through what we call the rulebook linter — it checks proposed content for within-envelope language before publish. The linter has rule packs for ABA 7.1, the advertising provisions of all 50 states, AICPA ET §1.400, and FCA SUP 15. The output is not a compliance certificate — it's a pre-flight check that flags anything a reviewing partner needs to substantiate or rephrase before the content ships.
The same pattern works without our software. A senior associate, an Excel sheet, and a habit of recording the substantiation source and the paraphrase envelope on every claim before publication gets you 80% of the protection. The remaining 20% is what we built the linter for.
Add a public "Substantiation Note" to claims that are likely to get paraphrased. A footnote of the form "Source: NLJ 250 (2025 ed.), filed cases through Q4 2025" attached to your "30+ ITC litigations" line does two things: it puts the substantiation in the source content where the model can see it (reducing out-of-envelope paraphrase risk), and it documents your good faith if a complaint ever arises. The cost is a tasteful footnote.
Rule 7.1 was not written for the LLM era, but its standard — truthful, substantiated, non-misleading — survives that era cleanly. The work is to operationalize it for content the model will re-render, not just content you publish directly.
The free 24-hour audit shows you specifically how the eight engines describe your firm against 200 high-intent legal and compliance queries.