What happened
In the insolvency case Jahani v Qiu (2026) FCA 1419, the plaintiffs employed to extract and summarise information from over 7,500 documents. After serving the AI‑assisted summaries on 11 August 2026, they issued a corrected version on 14 August, which the defendants later found contained errors. The defendants argued they lacked sufficient time to verify the accuracy of the summaries before the scheduled hearing on 28 September 2026. The Federal Court responded by vacating that hearing, resetting the timetable, and relisting the matter for 1 February 2027. The court also rejected the defendants’ request for a costs order based on the Federal Court’s Use of Generative Artificial Intelligence Practice Note (GPN‑AI), stating that the note does not penalise responsible AI use when accompanied by proper verification and human oversight.
The plaintiffs in Jahani v Qiu used a system—referred to in the judgment as “super intelligence”—to summarise over 7,500 documents related to alleged mis‑use of purchaser deposits exceeding $288 million.
After the initial AI‑generated summaries were served on 11 August, the plaintiffs manually reviewed and corrected the output, issuing an updated set on 14 August. Defendants’ spot‑checks identified errors, which the plaintiffs acknowledged but contested as immaterial.
Defendants argued that the timing of the AI‑assisted disclosures left them insufficient opportunity to verify the summaries before the hearing scheduled for 28 September. The Federal Court agreed, vacating the hearing and resetting the timetable to 1 February 2027.
The court declined to impose a costs order under GPN‑AI, explaining that the practice note is intended to encourage responsible AI use rather than punish it, provided the output is adequately verified and disclosed.
Source details: thelawyermag.com ↗
Why it matters
The decision clarifies how Australian courts may treat tools in large‑scale document review. By emphasizing that AI‑generated material is permissible provided it is transparently disclosed and verified, the ruling sets a practical precedent for litigants seeking to manage voluminous evidence efficiently. It also signals that cost sanctions will not automatically follow merely because AI was used, reducing uncertainty for parties considering AI‑assisted workflows. This guidance may influence broader procedural rules, encouraging courts to adopt technology‑friendly practices while safeguarding procedural fairness. The case also highlights the need for parties to coordinate timelines when AI tools are involved, as late disclosure can still trigger cost consequences and hearing delays.
The judgment provides concrete judicial guidance on the acceptable use of in large‑scale document review, a growing practice in complex litigation.
By rejecting a costs penalty solely based on AI use, the court reduces the financial risk for parties that adopt AI tools, potentially accelerating the adoption of AI‑driven legal tech.
The decision underscores the importance of procedural fairness: parties must disclose AI‑generated material early enough for opponents to verify, reinforcing transparency standards.
The case may influence future updates to the Federal Court’s practice notes, prompting clearer rules on verification, traceability, and human oversight for AI‑generated evidence.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
Impossibility results in algorithmic fairness (e.g. Kleinberg et al., Chouldechova) show what?
What to watch next
Future Federal Court practice notes may refine the standards for AI verification and disclosure, especially as generative models become more prevalent in litigation. Parties should monitor any subsequent rulings that address cost penalties for inaccurate AI outputs. Additionally, the adoption of AI by defendants to expedite their own review could lead to competitive pressures for more sophisticated document‑analysis tools, potentially spurring market growth for legal‑tech providers. Watch for any legislative or regulatory responses that codify AI use standards in Australian courts.
Potential revisions to the Federal Court’s GPN‑AI or new practice notes that detail verification protocols for AI‑generated summaries.
Subsequent cases that test the boundaries of cost sanctions when AI outputs contain material errors.
Legislative initiatives that could formalise AI disclosure requirements in Australian civil procedure.
Growth of legal‑tech vendors offering AI‑assisted document review platforms tailored to meet the court’s verification expectations.