If an AI tool can put a fabricated case in front of a judge, it can put a rule-breaking claim on your website. In August 2026 a lawyer in England was struck off after filing AI-generated authorities that did not exist — and then using AI again after the errors were flagged. The same month, the Solicitors Regulation Authority said it had logged 42 reports of AI misuse in a single year. The lesson for a law firm hiring an AI-marketing or GEO agency is direct: your website is lawyer advertising, an agency running AI-assisted content at volume will produce rule violations unless someone is screening every output, and the first thing to ask a prospective vendor is how they keep that from happening. This is a buyer’s guide to that one question — use it on us too.
What does a struck-off UK lawyer have to do with my marketing?
The mechanism is the same one that threatens your advertising. Generative AI produces fluent, confident text that is sometimes wrong — a fake citation in a brief, an invented statistic in a blog post, an implied guarantee in a practice-area page. The UK disciplinary case (SRA v Kumar, the first such matter before the Solicitors Disciplinary Tribunal) turned on unverified AI output reaching an official channel. Your marketing is a different official channel: every public word about your services is regulated communication under ABA Model Rule 7.1 and your state bar’s equivalent. An agency that pushes AI-written pages live without a human screen is running the identical risk on your letterhead — not fabricated cases, but unjustified-expectation language, unearned “specialist” claims, and dropped disclaimers.
Isn’t this just a UK problem?
No. The UK cases are the vivid version because the tribunal writes them down, but the underlying duty is universal. In the United States, Rule 7.1 bars any communication about a lawyer’s services that is false or misleading, including anything that creates an unjustified expectation about results. Courts across US jurisdictions have already sanctioned lawyers for AI-fabricated citations in filings, and state bars have issued guidance on the duty of competence when using AI. Advertising sits under the same umbrella. The point is not the country; it is that AI output is only as safe as the human review standing behind it — in a filing, and on a marketing page.
What are the top firms actually doing about it?
Building the human judgment in, not banning the tool. In September 2026 Sidley Austin became the first firm in London to make an AI rotation a compulsory part of its trainee training contract — a working seat where new lawyers do real assignments and are expected to assess AI output critically, spot its limitations, and judge when a human has to step in. That is the whole model in one sentence: AI to do the work, a trained human accountable for whether it is right. A serious AI-marketing vendor should mirror it — AI to produce content at useful volume, a human who knows lawyer-advertising rules accountable for every page before it publishes. “AI-powered, human-accountable” is not a slogan; it is the only defensible way to run regulated content through a model.
What should I ask a GEO or AI-marketing agency before I hire them?
Ask who screens the output, against which rules, and when. The answer you want is specific: every piece is screened against Rule 7.1 and your state’s advertising rules before it goes live, by someone who can name the failure modes — outcome guarantees, unsubstantiated superlatives like “best” or “#1,” the “specialist” and “expert” restrictions, missing “attorney advertising” and “not legal advice” disclaimers, and testimonials that imply similar results for others. An agency that treats compliance as your problem, or that can’t describe its screen, is telling you it ships unreviewed AI copy. In a regulated profession that is not a small gap; it is the gap.
Doesn’t leading with compliance mean weaker, more cautious marketing?
The opposite, in AI search specifically. The pages AI engines cite are the accurate, well-sourced, genuinely useful ones — the same qualities a compliance screen enforces. Outcome-promise language and empty superlatives are exactly what a careful review strips out, and exactly what makes copy read as untrustworthy to a model deciding whom to name. Compliance discipline and citation-worthiness are the same discipline pointed at two audiences: the bar and the engine. A vendor that leads with compliance is not slowing your marketing down; it is building the only kind of content that both stays inside the rules and earns the citation.
How this maps to choosing a vendor
The through-line is simple: AI raises the volume of content you can produce and the volume of ways that content can break a rule, so the vendor’s value is in the human accountability layer, not the generation. Ask any agency — including Legal Torch — to show you its compliance screen and name who signs off before anything publishes. For the full set of questions to put to a GEO agency, see how to choose a GEO agency for your law firm; for the rule itself, AI search and ABA Rule 7.1. If you want to see where your firm stands across the AI engines before you hire anyone, the free AI Visibility Audit is a 60-second start, or book a call. We measure visibility, and we screen every word — outcomes are never guaranteed.
