Yes, with two conditions that have nothing to do with writing quality. The page has to add something a reader could not get elsewhere, and every claim on it has to be one you can prove.
Key takeaways
AI can write publishable content for a professional firm, and Google’s published policy does not say otherwise. What it prohibits is scale without value. The practical test is whether each page would still be worth publishing if search did not exist — whether it answers a question a client actually asks, in a way that reflects what your firm knows.
For a regulated practice a second test applies and it is stricter: the page is marketing, and your regulator has a rule about marketing. That rule usually reaches accuracy and verifiability, which is precisely where generated copy is weakest.
Google’s spam policies for Google Web Search define scaled content abuse as “when many pages are generated for the primary purpose of manipulating search rankings and not helping users”, typically focused on creating large amounts of unoriginal content that provides little to no value, “no matter how it’s created”. The listed examples include using generative AI tools to generate many pages without adding value for users, stitching or combining content from different web pages without adding value, and creating many pages whose content makes little sense to a reader but contains search keywords.
Two things follow. The mechanism is irrelevant — a hundred thin pages written by people is the same violation. And “adding value” is doing the work in that sentence, which for a professional firm means the specific thing you have that a general article does not: the Canadian rule, the provincial variation, the recurring client problem, the number you can source.
The same document defines site reputation abuse, where third-party content is published on a host site mainly because of the host’s already-established ranking signals. Firms that let an agency publish syndicated material under the firm’s name should read that section carefully.
For an Ontario lawyer, rule 4.2-1 permits marketing of legal services only if it is demonstrably true, accurate and verifiable; is neither misleading, confusing nor deceptive nor likely to mislead, confuse or deceive; and is in the best interests of the public and consistent with a high standard of professionalism. Rule 4.3-1 separately prohibits advertising that the lawyer is a specialist in a field unless certified by the Law Society — a rule that a generated page describing the firm as “specialists in” something will breach without anyone intending it.
CPAs, engineers and insurance brokers work under their own regulator’s version. The common shape is the same: verifiable, not misleading, no implied credential you do not hold. Read yours before the content calendar is signed off, not after.
A firm website promotes a business interest, so section 74.01 of the Competition Act applies. Paragraph (1)(a) reaches a representation to the public that is false or misleading in a material respect. Paragraph (1)(b) reaches a statement, warranty or guarantee of performance or efficacy not based on an adequate and proper test, and states that the proof of that test lies on the person making the representation. section 74.011 reaches a false or misleading representation in a locator, which is worth knowing if page titles and URLs are being generated at scale.
This is the concrete reason to run a claims register on generated content. A model produces “the leading firm for”, “guaranteed within 48 hours” and confident comparative statements as a matter of style. Each of those needs a source or it comes out. Treadstone Law’s notes on misleading advertising and on comparative advertising cover the ground.
A mid-sized Ontario firm is offered a programme of 200 search-optimised articles at a low per-page price. The sample reads fluently and says nothing that could not be said about any firm in any country.
The firm runs a smaller version instead. It lists the thirty questions clients actually ask, in the words clients use. For each, a partner records five minutes of answer; a model turns that into a draft; the drafter opens every citation and cuts anything the partner did not say. Three claims are removed at the register step and one page is dropped entirely because the honest answer was “it depends on your contract” and the page could not improve on that. Each published page names the partner who reviewed it.
Thirty pages ship over a quarter rather than two hundred in a month. Each names the province, cites something openable, and answers in the first two sentences.
Measure enquiries that reference a specific page, questions the intake team no longer has to answer from scratch, and pages a partner is willing to send to a client. Not word count, and not published volume.
This page is written for a firm that delivers work to a book of clients. If the question is really about the front desk — intake, scheduling, recall and reminders — that lives on the professional practice owners page. If it is about your own month-end, reconciliation and payables rather than what you deliver to clients, that is accounting automation, or bookkeeping automation for coding and receipt capture. The tooling overlaps; the risk almost never does.
Will Google penalise us for using AI?
Not for authorship. Google’s spam policies for Google Web Search target the pattern — many pages generated primarily to manipulate rankings and not to help users, however they were created. The exposure comes from scale without value, not from the tool.
Do we have to disclose that a page was AI-drafted?
No general Canadian rule requires it for ordinary marketing content. Naming the professional who reviewed the page does more for a reader’s trust than a disclaimer, and it is also what your regulator’s accuracy rule effectively presumes.
Can we let an agency publish under our name?
You can, and you remain responsible for it under your regulator’s marketing rule and under section 74.01 of the Competition Act. Read the site reputation abuse section of Google’s spam policies for Google Web Search as well before agreeing to syndicated material.
A 30-minute call is enough to tell you whether AI pays for itself here.