What was hallucinated

Fabricated: Case Law | Appellant's initial factum contained seven cited cases that could not be found (six purportedly ABCA decisions); court found they were likely fabricated and traced to use of a large language model; amended factum was permitted and costs referred.

Details

The appellant's original factum contained references to seven cases that could not be located (six allegedly decisions of this Court). Respondent flagged the issue; appellant's counsel ultimately acknowledged a contractor-drafted factum and that a large language model may have been used. The Court allowed an amended factum and reserved costs, warning that use of LLM without verification may attract costs, contempt proceedings, or Law Society referral.Monetary sanction was determined in a subsequent decision (available here).

Sanction teardown · CA Alberta, Canada · 2025-09-26

Reddy v Saroya

What happened

In CA Alberta, Canada, a filing relied on an unnamed/unconfirmed AI tool to help draft legal argument. The court identified the following problems with the citations in that filing:

Fabricated (Case Law)

Appellant's initial factum contained seven cited cases that could not be found (six purportedly ABCA decisions); court found they were likely fabricated and traced to use of a large language model; amended factum was permitted and costs referred.

Which AI tool

an unnamed/unconfirmed AI tool. Note: Charlotin's public database records tool attribution only where a court order, brief, or reporting on the matter states it explicitly; "unidentified" or "implied" means the record indicates AI use but does not name a specific product — we do not guess.

Outcome

Adverse Costs Order (monetary penalty: 17550 CAD.)

Additional detail

The appellant's original factum contained references to seven cases that could not be located (six allegedly decisions of this Court). Respondent flagged the issue; appellant's counsel ultimately acknowledged a contractor-drafted factum and that a large language model may have been used. The Court allowed an amended factum and reserved costs, warning that use of LLM without verification may attract costs, contempt proceedings, or Law Society referral.Monetary sanction was determined in a subsequent decision (available here).

How Citation Safe would have caught this

Citation Safe runs three deterministic layers before a brief is filed: (1) does the citation exist against CourtListener's database of published opinions, (2) if quoted, does that exact language appear in the source, (3) does the cited case actually support the proposition it is cited for. Fabricated case citations fail Layer 1. Fabricated or misattributed quotations fail Layer 2 even when the underlying case is real. Misrepresented holdings — a real case cited for a proposition it does not support — are the target of Layer 3. None of these checks involve asking another language model whether the citation looks right; they are lookups and text-matches against the actual source, which is why a hallucinated citation has to survive a direct lookup against the authoritative source — not another model's opinion — to earn a VERIFIED stamp; our measured false-verify rate is published live at /quality.

Check a brief before you file it → · See our live false-verify rate

Source: https://www.damiencharlotin.com/documents/844/Reddy_v._Saroya_Canada.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).