AI Data Broker for Law Firms: Legal Data Licensing in Canada
Your know-how may be an asset. Your client files never are.
As an AI data broker for law firms, Signal & Form helps Canadian firms license material they genuinely own, never client files. Legal data licensing for AI is the careful packaging of material a law firm genuinely owns, such as precedents, templates, and firm-authored commentary, for companies building legal AI tools. Client files are not part of it. We help Canadian firms separate what might be licensable from what never is, and route clean material to the right buyers, with your own lawyers making every call on privilege and confidentiality.
You have years of precedents and know-how. You also have a duty of confidentiality that never expires.
Legal AI tools are showing up in every practice, and partners are starting to ask whether the firm's own work product is worth something beyond billable hours. Some firms have been approached by vendors asking about their documents. Others simply wonder why outside companies seem to profit from legal knowledge while the firms that create it do not.
The tension is obvious to any lawyer. Almost everything a firm holds arrived through a client relationship, and the duty of confidentiality survives the end of the retainer and continues indefinitely. So the starting assumption has to be that client files are off-limits, however well they are scrubbed.
What is left can still be valuable: material the firm created about the law, not about any client. Sorting the two takes a lawyer's judgment, and that judgment stays with your firm.
What Sells
Data in law firms & legal that AI buyers want
Precedents and clause banks, properly scrubbed
Forms, templates, and clause libraries drafted by the firm, stripped of anything traceable to a matter. Every document needs review by one of your lawyers, because a distinctive fact pattern can identify a client even with names removed.
Firm-authored commentary and know-how
Client bulletins, articles, CPD papers, practice guides, and research memos on points of law that contain no client facts. Expert explanation of how the law works is what general AI models lack.
Bilingual and civil law material
French and English drafting, and content on Quebec civil law. Public legislation is already bilingual, but well-drafted, parallel precedents and commentary written by practitioners are much rarer.
Training and evaluation material
Articling and associate training programs, model answers, and checklists. Lawyer-written questions and answers can help AI developers test whether a model reasons the way a competent lawyer would.
Who wants legal data, and what is actually scarce
Legal AI has drawn heavy investment from established publishers and AI labs. Thomson Reuters completed its acquisition of Casetext, the company behind the CoCounsel AI legal assistant, in August 2023. OpenAI has described working with Harvey on a custom-trained case law model after concluding that general models lacked the legal knowledge the work required.
Public case law is not scarce in Canada. CanLII publishes court and tribunal decisions free of charge, and even free sources come with terms: in 2024 CanLII sued Caseway AI in the BC Supreme Court, alleging bulk scraping in breach of its terms of use. The case settled in 2026 on confidential terms, with the allegations untested in court. What remains scarce is high-quality, expert-written, non-public material with clear ownership.
For most firms, the realistic route is through legal technology companies and established data channels rather than a direct pitch to an AI lab. Volume matters, so a first step is an honest inventory of how much clean, firm-owned material you actually have.
Before You Sell
Privilege, confidentiality, and ownership come first
- Privilege: the Supreme Court of Canada has said solicitor-client privilege must remain as close to absolute as possible. Nothing privileged goes into a licensed dataset, de-identified or not.
- Confidentiality is broader than privilege: law society codes protect all information about a client's affairs acquired in the relationship, regardless of whether others share the knowledge. Removing names is not enough when the facts are distinctive.
- Law society AI guidance: the Law Society of BC's generative AI guidance says client confidential information should ideally be omitted from anything supplied to an AI tool, and warns that tools may reuse information in ways you cannot control. The Law Society of Ontario's 2024 white paper flags the unanticipated spread of confidential information as a key risk. Licensing material for training goes further than using a tool, so treat this guidance as a floor and talk to your law society's practice advisors.
- Client consent is a narrow door: law society guidance sets a high bar for fully informed consent, and most firms will not want to ask clients for this. Plan as if consent is not available.
- Ownership: under the Copyright Act, work an employee creates in the course of employment generally belongs to the employer unless agreed otherwise. Partners are not employees, and contract lawyers, co-authors, and precedents adapted from publishers' form books may not be yours to license.
- Sensitive practice areas: family, criminal, immigration, and health-related files carry especially sensitive personal information. Keep them out entirely.
General information, not legal advice. Have a lawyer review rights, consent, and any licensing contract.
The Canadian angle
Canada's bilingual, bijural system is a genuine point of difference. Federal Acts are enacted in English and French, with both versions equally authoritative, and Quebec practises civil law alongside the common law provinces. Firms with strong parallel French and English drafting hold something most legal datasets lack.
Each provincial and territorial law society sets its own code of conduct, and the wording differs. Check your own code and your law society's AI guidance before any material leaves the firm.
Copyright is unsettled. Canada has no specific text and data mining exception, the federal government has consulted on copyright and AI, and a news publishers' lawsuit against OpenAI is proceeding in Ontario. Buyers want clear chain of title, and a firm that can document it has an advantage.
How It Works
From archive to licensing deal
We review what you hold, estimate a realistic value, work with your lawyer on rights and consent, document the dataset, and route it to the right buyers. You pay a success fee only if a deal closes. See the full process on our AI data licensing page.
FAQ
Questions about AI data broker for law firms
Some documents, possibly. Firm-authored material about the law, such as scrubbed precedents, articles, and training content, can be licensable if the firm owns it. Anything from a client file, privileged or not, should be treated as off-limits. This is general information, not legal advice; your firm and your law society make the call.
We will not broker them. The duty of confidentiality covers information about a client's affairs even if others know it, and distinctive facts can identify a client after names are removed. The risk to clients and to the firm's licence is not worth it.
As of October 2026, we are not aware of a Canadian law society rule written specifically for licensing firm material to AI companies. Existing duties of confidentiality, privilege, and competence still apply, and generative AI guidance from the Law Society of BC and the Law Society of Ontario points toward keeping client information out of AI systems. Ask your law society's practice advisors before you start.
Work created by employees in the course of their jobs generally belongs to the firm unless an agreement says otherwise. Partners, contract lawyers, and outside co-authors are different, and partnership or engagement agreements may decide it. Confirm ownership before anything is licensed.
It depends on volume and uniqueness. A firm with a deep, well-maintained precedent bank, specialist commentary, or strong bilingual drafting has a better case than one whose know-how lives in client files. We give you an honest read on the first call.
No. We start with an inventory conversation about categories and volumes. Your lawyers review and select documents; we never need to see client files, and anything we handle is material your firm has already cleared.
Related Reading
AI for Law Firms: What Actually Works in 2026
How to Sell Your Data to AI Companies from Canada: Process, Buyers, and the Law
What AI Companies Pay for Training Data in 2026: Every Public Price, Sourced
Other Industries
Sources: Thomson Reuters, Casetext acquisition · OpenAI, Harvey customer story · Canadian Lawyer, CanLII settles with Caseway AI · Supreme Court of Canada, Lavallee (2002 SCC 61) · Law Society of BC, duty of confidentiality · Law Society of BC, professional responsibility and generative AI · Osler, Law Society of Ontario generative AI white paper · Copyright Act, section 13 · Official Languages Act, section 13 · ISED, consultation on copyright and generative AI · BNN Bloomberg, publishers v OpenAI proceeds in Ontario
Sitting on law firms & legal data? Find out what it is worth.
Book a free call. Tell us what you hold, and we will give you an honest read on whether it is worth licensing.