Legal Marketing

Engagement Letter AI Clauses: Template + Provincial Variations for Canadian Law Firms

15+ years in local marketing; Google Ads certified; Shopify Partner.

Picture this. You're a family law solo in Calgary. A new client signs their retainer Monday morning. By Wednesday, you've used Lexis+ AI to pull case law, ChatGPT to draft a first pass of a separation agreement checklist, and an AI intake chatbot caught their initial inquiry at 11pm on Sunday. All three touches were fine. All three probably should have been disclosed in the engagement letter. And if your current engagement letter ai language is a single sentence or missing entirely, you've got a compliance gap that's going to get bigger every quarter.

This article is the template. I'm going to give you sample engagement letter ai clauses, the provincial variations you need to know about (Ontario, BC, Alberta, Quebec), and a short framework for deciding how much disclosure is enough. What I'm NOT going to cover here is whether you should be using AI at all, or which tools are worth it. For the tool-by-tool breakdown, see our guide to legal AI tools for Canadian law firms. For the Law Society rules on AI-generated marketing content, see AI content and Law Society rules.

One more thing up top. I'm not your lawyer. You're the lawyer. This is a template and a framework, not legal advice. Run it past your firm's risk committee, your insurer (LawPRO in Ontario, CLIA in BC, ALIA in Alberta), and whichever provincial Law Society guidance applies to you.

Why AI Disclosure in Engagement Letters Is Suddenly a 2026 Thing

Two years ago, most firms didn't mention AI in the retainer at all. The Federation of Law Societies of Canada updated its Model Code in 2024, and provincial Law Societies have been rolling out AI guidance through 2025 and into 2026.

Here's the short version of what changed. Marketing claims made with AI assistance need disclosure to avoid running afoul of Rule 4.2-1 (marketing must be "demonstrably true, accurate and verifiable"). Client-facing AI use, like intake chatbots or AI-drafted client communications, is increasingly expected to be disclosed up front. And the confidentiality piece, which I think is the part most firms miss, got louder: any time client information touches a third-party AI tool, you need to have thought about where that data goes.

A lot of firms still treat this as optional. I think that's a mistake. The engagement letter is the cleanest place to handle it, because it's already the document where you set expectations about fees, scope, confidentiality, and file retention. Adding AI disclosure there is a five-minute edit once you have the language.

A quick pattern observation. In my experience working with professional services firms adopting new tools, the ones who front-load disclosure almost never hear from clients about it again. The ones who bury it, or skip it, end up having awkward conversations six months later when a client reads a news article and calls to ask "are you using ChatGPT on my file?" Front-loading isn't just compliant. It's operationally easier.

The Core AI Clause: Template You Can Actually Use

Here's a baseline clause that works as a starting point for most Canadian firms. Again, run it past your insurer and your Law Society guidance before you ship it.

Use of Artificial Intelligence Tools

In the course of providing legal services to you, our firm may use artificial intelligence (AI) tools to assist with tasks including, but not limited to: legal research, document review, drafting of internal memoranda, summarization of case law, and administrative tasks such as scheduling and intake.

You acknowledge and agree that:

Supervision. All AI-assisted work product is reviewed by a licensed lawyer before being used in your matter or communicated to you or any third party. AI tools do not replace the professional judgement of your lawyer.
Confidentiality. We use AI tools that meet our firm's confidentiality and data security standards. We will not input personally identifying information, privileged information, or details of your matter into any public or general-purpose AI tool (for example, the free versions of ChatGPT, Gemini, or Claude). Tools used on your matter either (a) operate on closed systems that do not train on your data, or (b) are used only with fully anonymized inputs.
Accuracy. AI tools can produce errors, including fabricated citations. We verify all AI-generated legal research, citations, and factual assertions before relying on them in your matter.
Billing. Time spent using AI tools is billed at the same hourly rate as the lawyer or staff member using them, unless otherwise agreed. AI does not reduce our fees automatically; it may, in some cases, reduce the total time required for a task.
Your right to opt out. You may request, in writing, that we not use AI tools on your matter. We will accommodate this request where reasonably possible. In some cases, opting out may increase the time and cost required to complete your work, and we will discuss this with you before proceeding.

That's the skeleton. Five points. Supervision, confidentiality, accuracy, billing, opt-out. If you add nothing else to your engagement letter, add those five.

Now, a few things about that clause worth flagging.

The opt-out language in point 5 is the one I get the most pushback on. Some lawyers don't want to offer it. My view, and this is a judgement call, is that offering it costs you nothing and signals that you're thinking about this seriously. Most clients won't exercise it. The ones who do were going to ask anyway.

The billing language in point 4 is where firms get into trouble. If you're using AI to draft a memo in 20 minutes that would have taken 2 hours, and you bill 2 hours, that's a problem under most provincial rules on reasonableness of fees. The FLSC Model Code Rule 3.6-1 requires fees to be "fair and reasonable." Billing pre-AI time for post-AI work isn't fair. Bill the actual time. If AI saves you time, the client gets the benefit, and you get a client who stays.

Provincial Variations: What Changes Depending on Where You Practise

Every province has a slightly different take. Here's what you need to adjust based on where your firm is regulated.

Ontario (LSO)

Ontario's Rules of Professional Conduct Rule 4.2-1 requires that marketing of legal services be "demonstrably true, accurate and verifiable." The LSO has been active on AI guidance. Their general position, as of 2025-2026 communications, is that technological competence (Rule 3.1-2 commentary) now includes understanding AI tools' capabilities and limits.

What to add for Ontario:

In your confidentiality paragraph, reference your firm's compliance with the LSO's guidance on technology and client confidentiality.
If you use an AI intake chatbot on your website, the chatbot itself should clearly identify as AI. The engagement letter should confirm that any information collected through intake AI was reviewed by a licensed lawyer before the retainer was offered.
Ontario also has the strictest stance on testimonials (effectively a ban under Rule 4.2 commentary). If you're using AI to generate any marketing copy at all, your engagement letter and your website need to line up. For more on that, see AI content and Law Society rules.

British Columbia (LSBC)

BC's Code Rule 4.1-1 on marketing is substantively similar to Ontario's but with more specific disclaimer requirements for paid advertising. LSBC has emphasized the duty of technological competence explicitly.

What to add for BC:

BC firms should pay close attention to the "supervision" paragraph. LSBC guidance has emphasized that delegating to AI is delegating, and Rule 6.1-1 on supervision of non-lawyers applies by analogy.
If you're in BC and you use a cloud-based AI tool, confirm data residency. BC has privacy law (PIPA) considerations on personal information leaving the province or the country. Your confidentiality paragraph can reference this.

Alberta (LSA)

Alberta's Rule 7.2 is, historically, a bit more permissive on advertising than Ontario or BC. That doesn't give Alberta firms a pass on AI, but it does mean the LSA has tended to issue guidance rather than hard rules.

What to add for Alberta:

Alberta firms can use slightly more direct language about AI capabilities in marketing, but the engagement letter itself should still hit the five points.
ALIA (Alberta Lawyers Indemnity Association) has been vocal about AI hallucination risk in litigation. If you do litigation work, strengthen the accuracy paragraph to explicitly reference citation verification.

Quebec (Barreau du Québec)

Quebec is the most distinct. The Code of Professional Conduct of Lawyers has specific provisions on dignity, client communication, and bilingual service requirements.

What to add for Quebec:

Your engagement letter AI clause must be available in French. If the client's contract language is French, the full clause needs French parity, not a summary.
Quebec's Law 25 (the modernized privacy law) is stricter than most of Canada. Any AI tool that processes personal information about Quebec residents needs to meet Law 25 requirements. Your confidentiality paragraph should reference Law 25 compliance explicitly.
The "opt-out" right maps well to Law 25's consent framework, so Quebec clients are more likely to expect and exercise it.

The Confidentiality Piece Most Firms Get Wrong

I want to spend a minute on this because it's where I see the biggest gap between what firms write in their engagement letter and what actually happens day-to-day.

If anyone at your firm, lawyer, paralegal, intake admin, pastes client information into the free version of ChatGPT, that information is now in OpenAI's systems and may be used for training. That's a confidentiality breach. Full stop. It doesn't matter what your engagement letter says.

The engagement letter clause is the contract with the client. The internal AI policy is what makes the clause true. You need both. For the internal policy side, see our sample law firm AI policy language. For deeper coverage of the intake chatbot piece specifically, AI intake chatbots and UPL risks goes into where that line sits.

A worked example on the cost side. Let's say your firm spends 40 hours a month on legal research across all matters. You add Lexis+ AI at roughly $200-$400 per lawyer per month (check your actual quote, pricing varies). If that tool cuts research time by 25%, you're saving 10 hours a month. At a $400 blended hourly rate, that's $4,000 of attorney time freed up.

Here's the catch most firms miss. That $4,000 doesn't go to the firm's bottom line if you're billing clients for the saved time. It goes to the client, under fair-fee rules. Your margin comes from redeploying those 10 hours to more matters, not from billing the old hours for the new work. The engagement letter's billing clause is what keeps that straight.

Rolling This Out: A Four-Week Plan

If you're reading this and realizing your current engagement letter doesn't have any AI language, here's how I'd actually deploy the change. This is tailored to a small-to-mid-sized Canadian firm (1-20 lawyers).

Week 1: Audit and decide. List every AI tool currently in use at the firm. Include the obvious ones (Lexis+, Westlaw Precision, ChatGPT) and the less-obvious ones (Grammarly, Microsoft Copilot inside Word, Otter.ai for transcription, any intake chatbot on your website). For each tool, note: who uses it, what client data touches it, and what the vendor's data handling policy says. This is a list, not a strategy yet.

Week 2: Draft and internal review. Take the template clause above, adapt it to your province, and add the specifics from your Week 1 audit. Circulate to all lawyers at the firm for comment. Get sign-off from whoever handles your risk (managing partner, or the partner designated for professional responsibility).

Week 3: Insurer and Law Society check. Send the draft clause to your insurer for comment. LawPRO in particular has been responsive on this. Check your Law Society's current AI guidance against the clause. Update based on feedback.

Week 4: Deploy and retrofit. Update your standard engagement letter template for all new matters starting next month. For existing matters, decide whether to issue a supplemental disclosure. My view: for long-running matters where your AI use has meaningfully changed (for example, you just added Lexis+ AI), send a short supplemental letter. For routine matters closing in the next 60 days, skip it. Document your reasoning in the file.

Pattern observation. Firms that do this rollout in a concentrated month tend to stick with the updated language. Firms that try to do it "eventually" tend to have three different versions of the clause floating around six months later, and nobody knows which one is current.

Red Flags in Clauses I've Seen Floating Around

A few engagement letter AI clauses I've seen that I'd push back on.

"We may use AI in our practice." That's it. That's the whole clause. This is worse than nothing, because it creates a paper trail that you disclosed without actually disclosing anything useful. Clients can't consent meaningfully to a sentence.

"AI tools we use are fully secure and confidential." Don't make absolute claims you can't back up. Under Rule 4.2-1 equivalents across Canada, you need to be able to verify the claim. "Meet our firm's confidentiality standards" is defensible. "Fully secure" is not.

"We pass AI cost savings on to the client in reduced fees." Only write this if you actually do it, and you can prove it. I've seen this in firm marketing materials more than in engagement letters, but when it shows up in the retainer, it's a trap. See our article on advertising AI-powered legal services for the brand-vs-compliance tradeoff on this one.

Any clause that waives the firm's liability for AI errors. Professional responsibility rules generally don't let you contract out of your supervisory duty. If the AI hallucinates a case citation and you file it, that's on you, not on the AI vendor, and not on the client who signed a waiver. For the underlying risk, see AI hallucinations in legal work.

When to Get Outside Help

If your firm has more than 10 lawyers, multi-jurisdictional practice, or you're using AI in high-stakes areas (litigation, immigration applications, M&A), get a professional-responsibility lawyer to review your clause. The cost is a few hours of their time. The cost of getting it wrong is a Law Society complaint, an insurance claim, or both.

If you're a solo or a small firm, the template above plus a quick check with your insurer is usually enough. Add the provincial adjustments. Move on.

The broader strategic question of whether to market your firm as "AI-enhanced" at all, which is a different conversation than disclosing AI use in the retainer, belongs on our parent guide on SEO marketing for lawyers and the dedicated piece on advertising AI-powered legal services. The engagement letter is backstage. Your marketing is front-of-house. Both need to line up.

Three Takeaways

Every Canadian law firm using AI should have an AI clause in the engagement letter covering supervision, confidentiality, accuracy, billing, and opt-out. Five points. Not optional anymore.
Provincial variations matter, especially for Quebec (Law 25 + bilingual parity) and BC (data residency). Ontario and Alberta are closer to the default template, but still need the Law Society-specific framing.
The engagement letter clause is only as true as your internal AI policy makes it. The document is the contract; the policy is the operations. You need both, and they need to match.