Your competitor's blog now publishes three times a week. Yours has not moved since 2019, and your marketing coordinator just proposed 'letting AI handle content' so you can catch up before the next intake cycle. Part of you wants to say yes. Part of you remembers the headline about the lawyers sanctioned for citing cases ChatGPT invented in a federal filing. Both instincts are correct, and the tension between them is not a content problem. It is a Strategy decision, sitting on the desk of the same person who is supposed to be billing hours this afternoon.
AI-generated content is permitted on law firm websites, but only when a licensed attorney reviews every factual and legal claim before publication, per ABA Model Rule 1.1, Comment 8. Skipping that review risks Google's helpful-content system penalties, per Google's Search Quality Rater Guidelines, and potential bar discipline for misleading statements.
The Hard Truth
Content and SEO look like a marketing decision. They are a Strategy decision — the same decision that determines who your firm serves, at what price, and why a prospective client should trust you over the three other firms running identical Google ads in the same zip code. The naive version of that decision assumed content had to be produced by a human billing hourly, which is why most firms either overpaid for too little volume or published nothing for years. AI changes the unit economics of a first draft. It does not change the unit economics of legal accuracy — a hallucinated citation costs the same in credibility whether a paralegal or a language model produced it, and per the sanctions order in Mata v. Avianca, Inc., No. 22-cv-1461 (S.D.N.Y. 2023), a federal court does not distinguish between the two when deciding whether Federal Rule of Civil Procedure 11(b) was violated. Here is the skeptic's real objection, stated on its own terms: if software writes the words, software is practicing law, and no vendor disclaimer changes that. The objection is right about the risk and wrong about the mechanism. AI-generated legal content is not unauthorized practice of law when a licensed attorney reviews, edits, and takes responsibility for every claim before a reader sees it — that review is exactly what ABA Model Rule 1.1, Comment 8 already requires of any technology a lawyer uses competently. The objection breaks down where it assumes review is optional. It is not. It is the entire compliance mechanism.
What Happens If You Wait
Delay has a specific cost, not an abstract one. Every month your content strategy stays frozen, Google's Helpful Content system — per Google's Search Quality Rater Guidelines and its E-E-A-T framework — keeps rewarding competitors already publishing reviewed, expert-attributed content at volume, and the ranking gap widens in a way that is expensive to close later. Separately, unreviewed AI-drafted content or filings carry real exposure: the Mata v. Avianca sanctions order fined the responsible attorneys for citing fabricated cases, and Judge Brantley Starr's standing order in the U.S. District Court for the Northern District of Texas, Dallas Division, now requires attorneys to certify whether a filing was drafted with AI assistance and that a human verified its accuracy. State bars are moving the same direction — the State Bar of California's COPRAC and the Florida Bar's ethics opinion process have both signaled that unreviewed AI content published under a firm's name can trigger discipline under advertising and competence rules. None of this requires you to stop using AI. It requires you to stop publishing it without a review record.
Step-by-Step Process
The Fifth Decision pillar that governs this is Witness — not Refusal, not Voice, not Veto. Witness is the signed audit chain: a record of who reviewed a piece of AI-drafted content, when, and against what source, before it went live or into a filing. Judge Starr's standing order is a Witness requirement dressed up as a courtroom rule — certify that a human checked the machine's work. Your content workflow needs the same mechanism, because the New York State Unified Court System's Advisory Council on AI and the Florida Bar's ethics opinion process are both signaling that 'we used a tool' will not be an adequate answer if a website claim turns out false or misleading under a rule modeled on ABA Model Rule 7.1. A workable sequence, built to satisfy the strictest state you practice in: One. Draft with AI, treating the output as a fast, unlicensed research assistant's first pass — never a finished, citable source. Two. Route every draft through a named, licensed attorney who checks each factual and legal claim against a primary source before publication, satisfying ABA Model Rule 1.1, Comment 8. Three. Log the reviewer's name and review date in your CMS or document management system — the Witness record a Rule 11(b) inquiry or bar complaint would ask you to produce. Four. Where your state bar's guidance (COPRAC, Florida Bar) recommends disclosure for AI-influenced legal advice, disclose it; requirements differ by state, so confirm current guidance before publishing. Five. For multi-state firms, localize reviewed content by jurisdiction rather than duplicating it across state pages, which avoids both duplicate-content SEO penalties and one-size-fits-all legal claims.
A Real-World Example
Consider a composite, not an actual client: a nine-attorney personal injury firm operating across two states hired a content vendor promising fifty AI-generated blog posts a month to build out its SEO footprint. Traffic rose for about six weeks. Then a Google core update tied to the Helpful Content system cut the firm's organic rankings by more than half, because the posts were templated across both states with only the city name swapped — thin, unreviewed, and indistinguishable from competitor pages built by the same vendor. The managing partner did not abandon AI. She dropped the vendor, kept the drafting tool, and routed every draft through a rotating pair of associates who fact-checked and signed off before publication. Volume fell from fifty posts a month to twelve. Rankings recovered over the following two quarters, and because every post now had a documented reviewer, the firm also had a ready answer the one time a referring attorney asked who had verified the legal claims on the site.
William J. Vasquez did not start as a business operator. He has a BS in Computer Science, fifteen years practicing law, an M.Div., and seven years in the Air Force — training in law and theology, with zero formal business-operating education. He learned Scaling Up-style discipline the hard way, running a real law firm through the years when he had no framework for the business side at all. HODOS is the system he built to be the one he wished he had had — including the review workflow underneath this piece.
Key Terms Explained
E-E-A-T: Google's Experience, Expertise, Authoritativeness, and Trust framework from its Search Quality Rater Guidelines, used to evaluate content quality signals, including legal content. Helpful Content System: Google's ranking system that demotes content produced primarily for search engines rather than for people. Hallucination: when an AI model generates plausible but false information, including fabricated case citations — the issue at the center of Mata v. Avianca, Inc. FRCP 11(b): the federal rule requiring an attorney to certify, by signing a filing, that factual contentions have evidentiary support. ABA Model Rule 1.1, Comment 8: the ethics comment establishing that competent representation requires understanding the benefits and risks of relevant technology, including AI drafting tools. ABA Model Rule 7.1: the ethics rule prohibiting false or misleading communications about a lawyer's services. UPL (Unauthorized Practice of Law): providing legal advice without a license; a risk for AI content only when no licensed attorney reviews or takes responsibility for the output. Standing Order (AI disclosure): a judge's local rule, such as Judge Brantley Starr's order in the N.D. Tex., requiring certification of whether AI was used to draft a filing and that a human verified it.
Frequently Asked Questions
Can my law firm publish AI-written blog posts without violating state bar advertising rules? Yes, in most states, provided a licensed attorney reviews the content for accuracy before publication and the firm avoids false or misleading claims under rules modeled on ABA Model Rule 7.1; several bars, including the State Bar of California's COPRAC, are actively developing AI-specific guidance, so confirm your jurisdiction's current position before relying on general practice. Will AI-written website content hurt my Google rankings? AI-generated content itself is not penalized by Google — content that is unhelpful, unreviewed, or templated across pages is, under Google's Helpful Content system and E-E-A-T guidelines; the determining factor is editorial quality and attorney oversight, not the drafting tool. Will AI eventually replace legal content writers, or lawyers, entirely? AI changes the economics of producing a first draft; it does not change who is legally responsible for the accuracy of what gets published or filed, which is why every serious AI-disclosure rule — from Judge Starr's standing order to state bar ethics opinions — is built around human verification, not human replacement. Do I need to disclose AI use to clients? Not universally required yet, but several state bars are moving toward disclosure expectations for AI-influenced legal advice; check your own state's current guidance, including the Florida Bar and New York's Advisory Council on AI, rather than assuming a uniform national standard.
Managing partners who move to an attorney-reviewed AI content workflow tend to describe the same shift: fewer emergency rewrites before a deadline, a clearer paper trail when a client or bar inquiry asks who verified a claim, and a marketing calendar that survives a bad month instead of collapsing into it. That is a pattern our team hears repeatedly from firms using HODOS's review-and-publish workflow — not a guarantee of a specific ranking position or case outcome, since no software or marketing vendor can promise either.
None of this requires an overhaul this week. It requires one meeting: pull whoever currently produces your firm's content, and ask them to name the last three posts published and who reviewed each one before it went live. If there is no clean answer, that is the gap — not a reason to stop using AI, a reason to build the Witness record around it before your next bar renewal or the next Google core update, whichever comes first.
If you want a second set of eyes on your firm's current content workflow — what is published, who reviewed it, and where the gap sits between your marketing vendor's promises and your bar's actual rules — that is a conversation our team has with managing partners regularly. Schedule a workflow review with HODOS and bring your last ten published posts; we will tell you, specifically, where the Witness record is solid and where it is not.
- Scaling Up for Law Firms: The Four Decisions Explained
- HODOS Attorney-Reviewed Content Workflow
- AI Ethics and Bar Compliance Resource Hub
- The Fifth Decision: Witness, Voice, Refusal, and Veto
- Schedule a Firm Operations Consultation