AI Hallucinated Cases in Canadian Courts | Courtready
Last Updated: April 16, 2026

When AI Invents the Law:
AI Hallucinated Cases in Canadian Courts

Courts and tribunals across Canada have identified non-existent case authorities being cited in legal proceedings. In many instances, the use of artificial intelligence tools has been identified as the source.

EN FR
LIVE Try CaseCheck by Courtready | Canadian legal citation verification tool.
Fictitious Citations
Decisions Affected
Courts & Tribunals
Other AI Misuse

Canada’s database of AI hallucinated cases and fictitious citations.

This database identifies specific fictitious citations that were cited as real authority and tracks cases where courts flagged the improper use of AI. Explore our Sanctions Database to see how courts have responded to fictitious citations.
Think you can spot a fake citation? Take our 3-round challenge. Most people can’t tell them apart.
Round 1 of 3
Which of these is a fake citation?
Now imagine being a judge. If you struggled to tell real citations from fake ones, consider what happens when AI-generated authorities are filed in court without verification. Once fake citations enter the system, trust in every citation erodes.
Important Disclaimer:

Unless otherwise specified, every citation listed below was described by the adjudicator as non-existent, fictitious, fabricated, or otherwise not locatable in reported case law. Citations marked with an asterisk (*) indicate that, although the adjudicator found that the cited case could not be located or does not exist, the decision does not attribute the issue to the use of artificial intelligence or make any finding about how the citation was generated.

We do not take a position on how or why (i) any citation came to be included in a filing or (ii) AI was used. We also do not attribute intent, motive, or wrongdoing to any party beyond what is expressly stated in the decision itself.

Geographic Spread

AI Hallucinations and Fictitious Citations Across Canada

Click a province or territory to filter the database below.

Fewer More

Fictitious CitationAppeared InDateCourt / Tribunal
?

What about the AI-hallucinated citations
that weren’t caught?

This database only includes fictitious citations that adjudicators identified and flagged. For every case caught, others may have gone undetected. The true number of fake cases in Canadian courts that slipped through remains unknown.

Key Findings
This Week Last Week
82%
113 of 138 Decisions involving self-represented parties who submitted non-existent case authorities
74%
102 of 138 Decisions where the court or tribunal presumed or found that AI tools generated the non-existent cases
Tom Macintosh Zheng

Using AI to prepare legal documents isn’t inherently wrong. For many self-represented Canadians, it may be their only realistic and cost-effective path to justice. The problem isn’t that people are using these tools. The problem is that our courts and tribunals don’t yet have adequate safeguards to catch when these tools hallucinate.

Tom Macintosh Zheng Co-founder of Courtready
Courtready
Live

CaseCheck by Courtready

Don’t let fake cases become real law. CaseCheck is a Canadian legal citation verification tool that cross-references authorities against Canadian case law database. We help self-represented litigants and legal professionals identify AI-hallucinated cases before they reach the courtroom.

Check Out CaseCheck
AI hallucinations in Canadian courts
Understanding the Problem

Why Does AI Hallucinate Legal Citations?

Large language models like ChatGPT don’t retrieve information the way a search engine or legal database does. They generate text, one word at a time, by predicting what comes next. The result: fake cases in Canadian courts that look real but never existed.

1

AI predicts words, not facts

Large language models work by predicting the most probable next word in a sequence. They don’t look up cases in a database. When asked for legal authority, the model generates text that looks like a citation: a plausible party name, a realistic court abbreviation, a convincing year. None of it is verified against actual case law.

2

Legal citations are especially vulnerable

Case citations follow rigid, predictable formats: a party name, a year, a court code, and a number. This structure makes them easy for AI to mimic convincingly. A hallucinated citation like Smith v. Jones, 2019 ONCA 412 looks identical to a real one, and there is no visual tell that it’s fabricated.

3

The model doesn’t know it’s wrong

AI doesn’t distinguish between generating a real citation and a fictitious one. It has no concept of truth or accuracy, only statistical probability. The model produces what looks right with the same confidence whether the case exists or not. That’s what makes these errors so dangerous in legal settings where every authority must be verifiable.

This is a simplified explanation. AI hallucinations are an active area of research with no single agreed-upon cause or solution.

Methodology

How We Search

We have compiled this database through systematic and manual searches of reported decisions published by the Canadian Legal Information Institute (CanLII). We run weekly searches across all Canadian courts and tribunals using two categories of search terms.

The first category targets judicial language indicating that a cited authority could not be verified. These searches use proximity operators to identify decisions where terms such as “cannot find,” “unable to locate,” or “unable to find” appear near words like “case,” “citation,” “authority,” or “decision.” These keywords capture decisions where an adjudicator has noted that a cited case does not appear to exist, regardless of whether the adjudicator attributed the issue to artificial intelligence.

The second category targets decisions that explicitly discuss the use of artificial intelligence in legal proceedings, using terms such as “artificial intelligence,” “generative AI,” and “gen-ai.”

How We Verify

The decisions responsive to the search terms are reviewed individually to confirm the presence of fictitious citations. Relevant data that make their way onto the database are extracted directly from the decision itself. A PDF of the decision is next to each entry for transparency and public verification.

Coverage & Limitations

We have designed this methodology to be comprehensive within reported decisions published on CanLII. Any case in which a Canadian court or tribunal identified a fictitious citation using language captured by our search terms will surface in our results and be published in our database. We record each decision as a separate entry. Where the same parties appear in multiple proceedings, we count each decision independently because each one reflects a distinct judicial finding.

This database does not represent an exhaustive record of all fictitious citations in Canadian legal proceedings. Our research does not extend to decisions before January 1, 2024. Lastly, this database documents judicial findings. We do not take a position on how or why any citation came to be included in a filing, nor do we draw any inference about the individuals involved beyond what is expressly stated in the decision itself.

How to Cite This Database

If you cite this database in academic research, court filings, or news articles, three citation formats are provided below. Click “Copy” next to any format to copy it to your clipboard.

McGill Guide 9th ed.

Tom Macintosh Zheng, AI Hallucinated Cases in Canadian Courts: Database of Fictitious Citations (Toronto: Courtready, 2026), online: <https://courtready.ca/fictitious-citations-in-canadian-courts/> [accessed ].

APA 7th ed.

Zheng, T. M. (2026). AI hallucinated cases in Canadian courts: Database of fictitious citations. Courtready. https://courtready.ca/fictitious-citations-in-canadian-courts/

BibTeX
@misc{zheng2026hallucinated,
  author    = {Zheng, Tom Macintosh},
  title     = {{AI Hallucinated Cases in Canadian Courts: Database of Fictitious Citations}},
  year      = {2026},
  publisher = {Courtready},
  address   = {Toronto},
  url       = {https://courtready.ca/fictitious-citations-in-canadian-courts/},
  note      = {Accessed: }
}

If you cite this database in a court filing, law review article, news story, or research paper, we’d love to hear about it. Email tom [at] courtready.ca.

This research is led by Tom Macintosh Zheng, former Toronto-based commercial litigator and co-founder of Courtready. Read our press release and research summary.

With thanks to Damien Charlotin, whose global AI Hallucination Cases Database remains the definitive international tracker and includes Canadian decisions surfaced through this research.

If you believe any entry in this database contains an error, please contact us at admin [at] courtready.ca. We are committed to accuracy and will review and address any concerns promptly.