Sanction teardown · Supreme Court of Victoria, Australia · 2025-08-14
Director of Public Prosecutions v GR
What happened
In Supreme Court of Victoria, Australia, 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)Initial joint submissions cited non-existent Supreme Court of Victoria cases; court could not locate them and senior counsel admitted "They do not exist."
-
Fabricated (Case Law)Initial submissions included purported direct quotes from cases said to be decisions of this Court, which could not be verified and were withdrawn.
-
False Quotes (Doctrinal Work)Initial submissions contained fictitious quotes attributed to the Second Reading Speech for the 1997 Bill; these were removed in the revisions.
-
False Quotes (Doctrinal Work)Initial submissions contained fictitious quotes attributed to the VLRC 2014 Final Report; these were removed in the revisions.
-
Fabricated (Legal Norm)Revised submissions referred to legislation that does not exist.
-
Fabricated (Legal Norm)Revised submissions asserted a provision of the Act had been inserted and repealed, but the provision never existed.
-
Misrepresented (Case Law)Initial submissions also relied on incorrectly cited and inapplicable cases.
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
N/A
Additional detail
The court identified issues with the use of artificial intelligence in preparing written submissions. The submissions contained fabricated citations and fictitious quotes, which were initially filed as joint submissions by the defense and prosecution. Upon discovery, the defense counsel took responsibility, citing the use of AI without proper verification. The court allowed revised submissions to be filed, emphasizing the importance of accuracy in legal documents and the responsible use of AI. No professional sanctions or monetary penalties were imposed, but the court reiterated the need for adherence to guidelines on AI use in litigation.
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/664/Director_of_Public_Prosecutions_v_GR_2025_VSC_490_14_August_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).