A weekly market update, a check-in after a showing, a note when a listing gets a price change — AI drafts these quickly and the wording is usually fine. Where an agent gets caught out is not the wording. It is whether the message is being sent as a personal update to one client, or as a commercial message to a list, because Canadian law treats those two very differently.
Key takeaways
An AI-drafted update to a current client — a note that today’s showing went well, or that a financing condition clears Thursday — is personal correspondence about an existing transaction, and Canada’s anti-spam law is not built around that kind of message. The picture changes the moment the same drafting habit gets applied to a market update sent to a list of past clients, sphere-of-influence contacts, or anyone who has not recently transacted with you. That is a commercial electronic message under the Act, whether an AI tool wrote it or you typed every word yourself.
The statute’s core prohibition is direct: “it is prohibited to send… a commercial electronic message unless (a) the person to whom the message is sent has consented to receiving it, whether the consent is express or implied… and (b) the message complies with subsection (2).” Subsection (2) sets three structural requirements — the message must identify who sent it, provide a way to contact that sender, and carry “an unsubscribe mechanism.” The contact information has to stay usable for a real window afterward: the sender “must ensure that the contact information… is valid for a minimum of 60 days after the message has been sent.” None of this is about the quality of the writing. An AI tool can produce beautifully worded market commentary that still violates CASL if it goes to a list without the right consent basis and mechanics attached.
The Act treats the unsubscribe link as load-bearing, not decorative — and an AI drafting tool has no way to know whether the unsubscribe mechanism on your actual email platform functions correctly, because that is not a wording problem, it is a plumbing one. Test the unsubscribe link on any templated client update the same way you would test any other piece of client-facing infrastructure. A message that reads as fully CASL-compliant, with an unsubscribe line included, still fails the Act if clicking that line does not actually remove the recipient.
Once the sending mechanics are handled, the REALTOR® Code’s content rules still govern what the update actually says. Article 3.1 requires that “a REALTOR® shall fully disclose to his or her Client at the earliest opportunity any information that relates to the transaction,” and Article 3.2 bars intentionally misleading anyone “as to any matters pertaining to a property.” An AI draft that smooths over a delay, softens a financing problem, or restates a market shift more optimistically than the data supports is a disclosure and accuracy problem the tool cannot be trusted to catch on its own — read every client-facing draft against what you actually know before it goes out, especially anything the model summarized or paraphrased from a longer source document.
The cleanest way to use AI for client updates is to separate the two decisions it tends to blur together. Let the tool handle the drafting task it is good at — turning your bullet points about a showing, a price change, or a market shift into a clear, well-organized message. Keep the sending decision entirely separate and entirely yours: who is actually on this list, when did each of them last give consent to receive commercial messages from you, and does the platform’s unsubscribe mechanism on this particular template still work. An AI tool asked to send a market update to a client list will do exactly that if it is wired into your email platform — it has no way to check your CASL consent records unless you have specifically built that check into the workflow. Treat every AI-assisted send the way you would treat a mail-merge: the personalization is only as good as the list and permissions behind it.
CASL does not require every recipient to have filled out an opt-in form — the Act allows “express or implied” consent, and s.6(5) carves out specific circumstances, including a message “sent to a person who is engaged in a commercial activity and consists solely of an inquiry or application related to that activity.” That kind of narrow exception is exactly the sort of legal judgment an AI drafting tool has no visibility into: it cannot tell from a contact list alone whether a given name gave express consent last month, has an active implied-consent window from a past transaction, or should never have been added to a marketing list at all. Keep your own consent record — even a simple dated log of when and how each contact agreed to receive updates — separate from whatever tool drafts the message itself, and check a name against that record before an AI-assisted send goes out, not after.
The unsubscribe mechanism runs on its own clock too: once a recipient uses it, CASL requires that “effect is given to an indication… without delay, and in any event no later than 10 business days after the indication has been sent, without any further action being required on the part of the person who so indicated.” Treadstone Law’s guide to CASL’s unsubscribe requirements covers the mechanics.
Related: US terminology that leaks into AI drafts, the glossary entry on the unsubscribe mechanism, and the case file on a newsletter sent to a stale list.
Yes — the Act covers “electronic address,” which the regulations define broadly enough to include text messaging accounts alongside email. Apply the same consent and identification thinking to an AI-drafted SMS blast as to an email one.
A reply can support implied consent depending on the circumstances, but implied consent under CASL is not indefinite and the exact rules depend on the nature of the prior relationship. Keep a record of when and how consent was established rather than assuming a single reply settles the question permanently.
Be cautious — building a send list this way can pull in contacts you never had consent to email commercially, and PIPEDA’s purpose-limitation principle expects personal information to be used only for purposes a reasonable person would consider appropriate in the circumstances it was collected.
A short call is enough to map where AI tools help your practice and where a human read-through still has to happen.
