This page is general information, not legal or ethics advice. Attorney Pro Marketing is a marketing agency, not a law firm. Rules of professional conduct differ between jurisdictions and guidance on AI is developing quickly. Consult your state bar's ethics counsel on anything touching client communications, confidentiality or advertising review.
Two things are true at once. AI has become genuinely useful for the mechanical parts of law firm marketing, and the places lawyers are most tempted to use it — publishing content under their name, and putting it between a distressed caller and the firm — are the two places where the downside is real.
- AI-written content is not penalised for being AI-written. It is penalised for being unhelpful, which it often is when unedited.
- Published content under your name is advertising, and your bar's rules apply to it whoever typed it.
- An AI receptionist is useful after hours and risky as the front door.
- Confidentiality obligations do not pause because a tool is convenient.
- The safe uses are the unglamorous ones.
Does AI-Written Content Hurt Your Rankings?
The short answer is that search engines have been clear they reward helpful, reliable content and do not treat automation as disqualifying in itself. The practical answer is that most unedited AI output is mediocre, and mediocre content performs badly — so the observed correlation between AI content and poor results is real even though the stated cause is not.
For law firms there is a sharper problem than ranking. Legal content that is subtly wrong is worse than no content. A generic article about bankruptcy exemptions that does not distinguish your state's rules is not a thin-content problem, it is a page that may mislead a reader who is about to make a decision about their house. The review burden on legal content is higher than on almost any other vertical, which erodes most of the time AI was supposed to save.
Where it does save time honestly: outlining, reorganising something you have already written, drafting the plain-language version of an explanation you gave verbally, and producing first drafts of the repetitive structural parts of a page. What it cannot do is supply the judgment that makes legal content worth reading, which is the part your prospective clients are actually responding to.
Where Marketing Content Meets the Advertising Rules
This is the part most discussion skips. Content published under your firm's name is advertising, and your jurisdiction's rules apply to it regardless of who or what drafted it. The failure modes are predictable:
- Unsubstantiated superlatives. Language models are trained on marketing copy and reach for “best”, “leading” and “top-rated” by default. Many bars restrict exactly these claims.
- Implied outcome guarantees. “We will get you the compensation you deserve” is the kind of sentence that appears unprompted and that advertising rules exist to prevent.
- Fabricated specifics. Invented statistics, invented case results and invented testimonials are a known failure mode of generative tools. Fabricated case results and testimonials are a serious advertising violation in every jurisdiction.
- Jurisdictional drift. Content that silently describes another state's law, on a page aimed at local clients.
- Missing disclaimers. Required disclaimers do not insert themselves.
The practical control is unglamorous: a lawyer reads every published word before it is published, with the advertising rules in mind, exactly as they would for a brochure. Our attorney advertising compliance guide covers the underlying rules.
AI Receptionists and the First Impression
AI phone answering has improved quickly and genuinely solves a real problem: a large share of firms cannot be reached by phone, and after-hours inquiries go unanswered. Capturing those is worth something.
The risk is specific to legal work. Someone calling a criminal defence or family lawyer is often in acute distress, and the moment they realise they are talking to software is a moment when some share of them hang up and call the next firm. That is a cost that does not appear in the vendor's capture-rate reporting, because a caller who hangs up was never captured.
A reasonable middle position: use it after hours rather than as the front door, make what it is apparent rather than letting the caller work it out, keep its job to capture and scheduling rather than anything resembling advice, and test it yourself by calling your own firm as a stranger would. Our page on law firm intake and missed calls covers the wider problem it is trying to solve.
One hard line: a tool that answers substantive legal questions for a caller is doing something a marketing vendor should not have configured and a firm should not permit.
Disclosure
Whether and when to tell clients that AI was involved is unsettled and jurisdiction-dependent, and it is exactly the kind of question your bar's ethics line will engage with. The distinction worth holding onto is between AI used on marketing — drafting a blog post, generating an image for a page — and AI used in client communications or legal work, where the considerations are materially different and more serious.
Separately from professional rules, consumer-facing honesty is its own argument. A client who discovers after the fact that what they took to be a personal response was automated has learned something about the firm that no amount of marketing recovers.
Confidentiality
The simplest rule, and the one most often broken casually: do not put client confidential information into a general-purpose AI tool. That includes the apparently harmless — pasting the facts of a live matter to generate a case study, uploading an intake form to summarise it, feeding a client email in to draft a reply.
Marketing creates its own version of this trap, because the most compelling content a firm could write is about its actual matters. A case study drawn from a real client is a confidentiality question before it is a marketing question, and the answer does not change because a tool made it easy to write.
Where AI Is Straightforwardly Useful
Without a meaningful compliance dimension: reformatting and reorganising your own existing material, drafting alt text and metadata, producing first-draft outlines you will rewrite, summarising your own analytics, generating variations of ad copy for a human to select from and check, and transcribing your own recorded explanations so they can become written pages. These are real savings and none of them involve a model's judgment reaching a client.
What This Means for Visibility
A connected question is whether AI assistants will recommend your firm at all — a different topic, covered on our page about AI search optimisation for law firms. Worth noting here only because the two get conflated: using AI to produce your content and being cited by AI are unrelated problems, and the second is not helped by the first.
Want a second read on your marketing content?
The free audit includes a compliance pass over your published pages — superlatives, implied guarantees, missing disclaimers — whoever or whatever drafted them.
Last reviewed October 2026. Guidance on AI is developing quickly; verify the current position with your own jurisdiction.
Frequently Asked Questions
Search engines have been clear that they reward helpful, reliable content and do not treat automation as disqualifying in itself. In practice most unedited AI output is mediocre and performs accordingly, so the correlation is real even though automation is not the stated cause. For law firms the sharper risk is accuracy: legal content that is subtly wrong, or that silently describes another state's law, is worse than no content.
Using it to draft and reorganise marketing material is widely regarded as unremarkable, provided a lawyer reviews everything published. The considerations become more serious where AI touches client communications, confidentiality or anything resembling legal advice. Guidance differs by jurisdiction and is developing, so this is a question worth putting to your state bar's ethics line rather than resolving from a blog post.
The predictable ones are unsubstantiated superlatives such as “best” or “leading”, implied outcome guarantees, fabricated statistics, case results or testimonials, content that describes another jurisdiction's law, and missing required disclaimers. Content published under a firm's name is advertising and the rules apply regardless of what drafted it.
It is genuinely useful for after-hours capture, scheduling and routing. The risk particular to legal work is that callers are often in distress, and some share will hang up and call another firm once they realise they are speaking to software — a cost that never appears in capture-rate reporting. Using it after hours rather than as the front door, keeping its role to capture and scheduling, and testing it by calling your own firm are reasonable precautions.
Client confidential information should not be entered into general-purpose AI tools, and marketing creates a particular temptation here because real matters make the most compelling content. A case study drawn from a real client raises a confidentiality question before it raises a marketing one, and ease of drafting does not change the analysis.
It is unsettled and jurisdiction-dependent, and worth asking your bar directly. The distinction that matters is between AI used on marketing material and AI used in client communications or legal work, where the considerations are materially more serious. Separately from any rule, a client who later discovers an apparently personal response was automated has learned something about the firm that marketing cannot undo.