An Australian court referred solicitors to a commissioner over AI submissions citing fake cases

In Pasuengos v Minister for Immigration and Citizenship (No 2), the Federal Circuit and Family Court of Australia found that a junior solicitor used a Google search combined with an AI summary to produce legal research containing three fabricated case citations, which were filed with the court without verification. The principal solicitor failed to independently check the authorities before they were submitted. Both solicitors were referred to the Legal Profession Conduct Commissioner (SA) and personally paid $3,125 in costs.

Anonymised law firm (names sealed by court order in Pasuengos v Minister) · Incident Feb 3, 2026 · Indexed Jun 4, 2026 · 3 sources

An AI summary confidently invented three legal authorities, and the human verification layer that should have caught them simply did not exist.
What
In Pasuengos v Minister for Immigration and Citizenship (No 2), the Federal Circuit and Family Court of Australia found that a junior solicitor used a Google search combined with an AI summary to produce legal research containing three fabricated case citations, which were filed with the court without verification.
Incident date
Feb 3, 2026
Who
Anonymised law firm (names sealed by court order in Pasuengos v Minister)
Failure mode
Hallucination
AI surface
Search / RAG
Severity
Medium

What happened

A junior solicitor with a restricted practising certificate at an anonymised law firm used a Google search combined with an AI summary to research legal authorities for submissions in an immigration case before the Federal Circuit and Family Court of Australia. The AI output included citations for three cases that do not exist, which were filed with the court on 8 May 2025. The principal solicitor who supervised the junior solicitor failed to independently verify the cited authorities against authoritative legal databases. When the respondent's counsel could not locate the cases, the solicitors conceded the citations were fabricated and admitted the submissions had been generated using AI without proper verification.

What broke inside the model

Failure path · mode profile · Hallucination
  1. 01 · TriggerA user asks for a fact, a citation, or a figure.
  2. 02 · Model stepThe model writes a fluent, confident answer.
  3. 03 · Control gapNothing ties the claim back to a real source.
  4. 04 · FailureA fabricated fact ships as if it were verified.
  5. 05 · ConsequenceThe false claim reaches a customer, a court, or the public.

Confidence holds, and even spikes, as the claim detaches from any source.

The AI summary tool generated plausible-sounding but entirely fictitious case citations, a well-known hallucination failure mode of large language models when queried about legal authorities. The supervisory safeguard that should have caught these fabrications failed because the principal solicitor did not independently verify the cited cases against authoritative legal databases. The system broke at two points: the AI producing fabricated outputs and the human review layer being absent.

Public visibilityMedium
Regulatory exposureActive
Customer impactFew customers
Financial impactEstimated
Time to disclosureMonths
  1. Court FilingPasuengos v Minister for Immigration and Citizenship (No 2) [2026] FedCFamC2G 96 (3 February 2026)austlii.edu.au
  2. PrimaryAI Hallucination Cases Database - Australia entriesdamiencharlotin.com
  3. PressAI Hallucination Cases: Australia - OBITERkirstenegroth.blog
Permalinkhttps://failureindex.ai/failures/australian-court-referred-solicitors
CitationAI Failure Index. "An Australian court referred solicitors to a commissioner over AI submissions citing fake cases" (FI-0136). Realm Labs. https://failureindex.ai/failures/australian-court-referred-solicitors (indexed Jun 4, 2026).
Share cardA branded image of this record for posts and slides.

Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0136. Full dataset at /data.

Note from Realm Labs, the Index steward

How Realm would have caught this

Controls for this failure mode
  • Prism
  • OmniGuard
  • AI Detection & Response (AIDR)

A runtime layer that watches the model's internal state can flag the moment a model commits to a claim it has no support for, and hold or reroute the response before it reaches a user. Realm reads those signals in real time rather than grading the transcript after the fact.