Zhang v. Chen and What It Means for Every Canadian Lawyer Using AI

In 2024, a lawyer in British Columbia filed submissions in a civil matter that cited cases from courts across Canada. The citations looked authoritative. The case names were plausible. The propositions they stood for were exactly what you would expect from a well-researched brief.

The problem: several of the cases did not exist.

The citations had been generated by an AI tool — and the lawyer had filed them without verifying them against any legal database.

The BC Supreme Court sanctioned the lawyer in Zhang v. Chen, 2024 BCSC [citation]. The decision joined a small but growing body of case law in which courts across North America have imposed sanctions for AI-generated fabrications.

What the court actually said

The court's concern was not that the lawyer had used AI. Canadian courts have not held that AI use is per se improper. The concern was the absence of independent verification.

A lawyer's duty of competence — codified in every provincial law society's rules of professional conduct — requires that lawyers know the law they cite and take responsibility for the accuracy of the materials they file. Using AI does not modify that obligation. Filing AI output without verification is filing unverified work product, whether the unverified source is AI or a junior associate.

Why this matters beyond one case

Zhang v. Chen was not unique. Kaur v. Desso (2025) followed the same pattern: AI-generated citations, no independent verification, sanctions. In the United States, Mata v. Avianca and its successors established the same principle. The pattern is consistent: courts will sanction lawyers who file AI content they have not independently verified.

The risk is not hypothetical. It is documented, repeating, and career-ending.

What verification actually requires

Canadian law society guidance converges on a consistent standard. Before filing any AI-generated research, a competent lawyer should:

1. 1.Verify every citation against an authoritative database (CanLII, Westlaw, LexisNexis)
2. 2.Confirm that the cited case stands for the proposition attributed to it
3. 3.Review adverse authority, not just supporting cases
4. 4.Apply independent professional judgment to the research
5. 5.Document the review process

The last point is underappreciated. Documented diligence — a record that you checked each citation, reviewed the full case, and applied your judgment — is your professional protection if a citation is later challenged.

How Juris addresses this directly

Juris's Sage research companion was built with Zhang v. Chen in mind. Every citation Sage produces is checked against CanLII before it reaches you. Sage cannot present a citation it has not verified. If a case does not exist, Sage tells you rather than fabricating it.

The verification workflow goes further: every piece of Sage output is watermarked [Unverified] until you complete a review checklist. The checklist mirrors the law society standard — check each citation, trace each factual claim, apply your professional judgment. Your completed review is logged with a timestamp and stored in your audit trail.

This is not a feature. It is the standard. Juris makes it faster, documented, and auditable.

The lesson

Zhang v. Chen is not a story about AI being dangerous. It is a story about verification being non-negotiable. Canadian courts and law societies have been consistent: you may use any research tool you choose, including AI. What you may not do is file unverified work product.

The lawyers who will use AI safely — and productively — are the ones who treat AI output the way they treat any research: as a starting point, not a conclusion.