Asif v Minister for Immigration and Citizenship [2026] FedCFamC2G 1402
Dismissing a student-visa judicial review, the Federal Circuit and Family Court of Australia fixed costs of AUD 9,097.93 against a self-represented applicant, increased because of AI misuse, after his ChatGPT-written submissions cited nonexistent paragraphs and cases. The judgment reproduces the judge's exchange with ChatGPT, in which the tool admitted having spun a fictional story about one of his judgments.
- Court
- Federal Circuit and Family Court of Australia (Division 2)
- Date
- 2026-07-22
- Who used the AI
- Litigant in person
- AI tool
- ChatGPT (admitted by the applicant)
- Conduct
- A self-represented student-visa applicant's submissions cited a nonexistent paragraph of Garikimukku, misused Inderjit, and cited Patel [2026] FedCFamC2G 426 for a proposition it did not decide (the court thought the reference itself was probably an unintended ChatGPT confusion with Patel 2019). He could not explain his own submissions at the hearing and admitted using ChatGPT.
- Outcome
- Costs order
- Penalty
- A$9,097.93
- Regulatory outcome
- No separate regulatory disposition recorded
- Status
- Verified against listed sources
- Last checked
- 2026-09-03
What was fabricated or misused
- Garikimukku v Minister for Immigration and Citizenship [2026] FedCFamC2G 9 at [37]–[39] — no paragraph [39], and Garikimukku did not support the claimed principles
- Eros v Minister for Immigration, Citizenship, Migrant Services and Multicultural Affairs [2020] FCA 1061 — cited through Garikimukku, but did not support the claimed principles
- Inderjit v Minister for Immigration, Citizenship, Migrant Services and Multicultural Affairs [2019] FCAFC 217; (2019) 272 FCR 528 at [41] — did not support the claimed principle
- Patel v Minister for Immigration and Citizenship [2026] FedCFamC2G 426 at [53], [55] and [58] — likely unintended; the cited paragraph concerned materiality, not cl 500.212 construction
What the court said
“At the Hearing, the Applicant was asked if he had used generative artificial intelligence (AI) to produce his submissions. The Applicant quite candidly told the Court he had used ChatGPT to assist with the submissions.”
“The dangers of AI in the legal area are well known. Its propensity to hallucinate and provide false references for propositions is well documented”
“I need to be completely upfront with you. I completely fabricated those details.”
“The Applicant’s late filing of AI-generated submissions, which he neither understood nor was able to articulate in Court, required the First Respondent to undertake additional work to provide supplementary submissions.”
Timeline
- Originating application for judicial review was filed
- Registrar made orders for staggered written submissions before hearing
- First Respondent filed submissions without the applicant’s material
- Applicant emailed Chambers his written submissions and authorities
- Hearing proceeded; applicant admitted using ChatGPT
- First Respondent filed supplementary submissions after late applicant material
- Court dismissed the application and fixed AUD 9,097.93 costs
Why this case matters
Asif extends the Australian migration line from Valu and JNE24 to a self-represented applicant. The court could not refer ChatGPT, OpenAI or the applicant to a regulator, so the consequence was costs and a warning that AI-written late submissions may be refused. The judgment also records the judge’s own ChatGPT exchange, making the risk concrete: the tool admitted inventing details about a judgment by the same judge. Unlike Valu, the source did not identify a lawyer with filing duties; it focused on cost consequences and court procedure.
Practice note
A self-represented migration applicant should file on time and use AI only for drafts checked against the tribunal reasons and AustLII authorities. Each cited paragraph must exist and support the stated proposition. If the applicant cannot explain a submission at the hearing, it is safer to withdraw or narrow it than to rely on a polished AI draft.
Primary sources for this incident
Related regulation records for this incident
No related regulation record has been linked to this incident yet.
Similar incidents on the record
- MFRD and National Disability Insurance Agency (NDIS), [2026] ARTA 18332026-08-26 · Administrative Review Tribunal of Australia · WarningIn Australia's Administrative Review Tribunal, a self-represented NDIS applicant filed submissions prepared with generative AI that contained fictitious tribunal decisions and inaccurate statutory references. General Me…
- Shahin v Industrial Relations Secretary on behalf of Multicultural NSW (No 2) [2026] NSWIRComm 112026-07-21 · Industrial Relations Commission of New South Wales (Full Bench) · OtherA Full Bench of the Industrial Relations Commission of New South Wales refused leave to appeal an unfair-dismissal decision, criticising the self-represented appellant's use of generative AI under Practice Note 33 after…
- JNE24 v Minister for Immigration and Citizenship [2025] FedCFamC2G 13142025-08-15 · Federal Circuit and Family Court of Australia (Division 2), General Federal Law · Costs orderIn Australia's Federal Circuit and Family Court, a Western Australian lawyer filed migration submissions citing four fake authorities produced using Claude and 'validated' with Microsoft Copilot, admitting overconfidenc…
- Neill-Fraser v Tasmania Parole Board (Supreme Court of Tasmania, reported 15 Sep 2026)2026-09-15 · Supreme Court of Tasmania · OtherIn Tasmania, ABC and the Human Rights Law Centre reported that Justice Stephen Estcourt set aside Susan Neill-Fraser's media-related parole conditions after the State conceded the Parole Board relied on a document with…
Cite this record
SafeLegalAI Global Legal AI Incident Tracker, “Asif v Minister for Immigration and Citizenship [2026] FedCFamC2G 1402”, safelegalai.com/tracker/asif-v-minister-immigration (accessed 2026-10-07). Data: CC BY 4.0.