Key Takeaways
- The Federal Court of Australia has issued a new practice note governing the use of generative artificial intelligence (AI) in legal proceedings.
- Chief Justice Debra Mortimer warns that presenting false or AI‑generated information to the court is “unacceptable” and can frustrate the just resolution of cases.
- Lawyers must verify that any AI‑generated citations, quotes, or facts are accurate and disclose the use of AI at the start of documents.
- Inserting confidential, suppressed, or private material into AI tools may lead to serious unintended consequences.
- Non‑compliance can result in adverse costs orders, sanctions, and breaches of legal and professional obligations.
- At least 73 Australian cases have already involved AI‑generated false citations; a Victorian lawyer was stripped of his right to practise as a principal lawyer for such misconduct.
- Judges are increasingly acting as “human filters” for AI‑produced legal arguments, a role Chief Justice Stephen Gageler described as reaching an “unsustainable phase.”
- The court embraces technology for efficiency but stresses that AI must be used appropriately and with due care to protect public confidence in the justice system.
Background: Rise of AI‑generated Errors in Court Filings
The Federal Court’s recent guidance comes amid a noticeable surge in court filings worldwide that contain false citations, fabricated quotations, or other inaccuracies traced back to generative AI tools. In Australia alone, regulators have identified at least 73 instances where AI‑produced content led to erroneous legal references. One notable example involved a full‑court judgment that noted, “We apprehend that the reference may be a product of hallucination by a large language model,” highlighting how AI can invent case law that never existed. This trend has raised alarms across the legal profession, prompting the Federal Court to act pre‑emptively before the problem compromises the integrity of proceedings.
Federal Court’s New Practice Note on AI Use
On Thursday, the Federal Court released a practice note that sets out clear parameters for how generative AI may be employed in litigation. The note applies to pleadings, written submissions, affidavits, expert reports, and any other documents lodged with the court. It emphasizes that while the court “embraces” technological advances that can increase efficiency, such tools must be deployed responsibly. The practice note is not a ban; rather, it establishes safeguards designed to curb the risks of AI hallucinations and ensure that any AI‑assisted work meets the same standards of accuracy expected from purely human‑crafted submissions.
Chief Justice Debra Mortimer’s Core Principles
Chief Justice Debra Mortimer articulated the court’s stance in unambiguous terms: “Presentation of false or inaccurate information to the court is ‘unacceptable.’” She added that such conduct “is inconsistent with the responsibility on all persons to not mislead the court or other parties” and is “likely to frustrate the just resolution of proceedings according to law and as quickly, inexpensively and efficiently as possible.” Mortimer’s language underscores that the duty of candor to the tribunal remains paramount, regardless of whether the source of the error is a human oversight or an AI model. Her statements serve as both a warning and a reminder of the ethical foundations undergirding Australian litigation.
Requirements for Disclosure and Verification
The practice note imposes concrete obligations on legal practitioners. Lawyers and solicitors must confirm whether AI was used in preparing any document, verify that all legal authorities cited actually exist, and ensure that those authorities support the propositions advanced. For affidavits and expert reports, if generative AI contributed to the drafting, the final product must still reflect the deponent’s or expert’s own recollection, knowledge, or experience. Crucially, the disclosure of AI use must appear “at the start of the document,” specifying where and how the technology was employed—whether for summarising information, analysing data, generating images, videos, sound recordings, or any other purpose that could affect evidence admissibility. This transparency allows the court to assess the reliability of the material and decide whether additional scrutiny is warranted.
Risks of Confidential Information Input
Mortimer also cautioned against feeding confidential, suppressed, or private information into generative AI systems. She warned, “There may be serious consequences for entering information into generative AI tools, even if sharing that information was not intended.” Once data enters an AI model, there is no guarantee it will remain secure; the model could inadvertently reproduce or expose sensitive details in subsequent outputs. Such breaches could violate client confidentiality obligations, privilege rules, or statutory suppression orders, exposing lawyers to disciplinary action and potentially undermining the fairness of the proceedings. The note therefore advises practitioners to treat AI tools as they would any other third‑party service handling privileged information—exercising extreme caution and, where feasible, avoiding the upload of sensitive material altogether.
Potential Sanctions and Professional Consequences
Failure to adhere to the new guidelines can trigger a range of sanctions. The chief justice noted that those who use generative AI in ways that conflict with the practice note “should expect consequences such as adverse costs orders and issues with compliance with legal and professional obligations.” In practical terms, a judge may order the offending party to pay the other side’s costs, strike out pleadings, or even refer the matter to a regulatory body for further investigation. Repeated violations could lead to findings of professional misconduct, fines, or, in severe cases, suspension or removal from practice. The Federal Court’s stance makes clear that the efficiency gains promised by AI cannot justify compromising the integrity of the judicial process.
Precedents and Past Sanctions in Australia
The court’s warnings are not theoretical; they reflect a growing pattern of enforcement. A Victorian lawyer became the first in the country to face sanctions after regulators discovered false citations in his filings generated by AI, resulting in the loss of his right to practise as a principal lawyer. Similar investigations have since been opened by legal authorities in Western Australia and New South Wales, indicating a coordinated national response. These cases demonstrate that regulators are willing to treat AI‑induced errors as serious breaches of duty, reinforcing the Federal Court’s message that ignorance of AI’s limitations will not be a defensible excuse.
Judicial Perspective: Judges as Human Filters
Chief Justice of the High Court Stephen Gageler offered a sobering assessment of the judiciary’s current role, stating that judges in Australia are increasingly acting as “human filters” for legal arguments created using AI. He described the situation as having reached an “unsustainable phase,” suggesting that the volume of AI‑generated content is outpacing the court’s ability to vet it efficiently. Gageler’s comment highlights a systemic strain: while AI can draft pleadings quickly, the judicial workload of verifying every citation and factual assertion may grow disproportionately, threatening timelines and increasing costs. His observation reinforces the need for proactive, upstream controls—such as those outlined in the practice note—to curb the flow of unreliable material before it reaches the bench.
Balancing Innovation with Judicial Integrity
The Federal Court’s approach seeks to strike a delicate balance between embracing technological innovation and safeguarding the core tenets of justice. By mandating disclosure, verification, and careful handling of confidential data, the court aims to harness AI’s potential for efficiency—such as expediting legal research or drafting routine documents—while preventing the erosion of trust that could arise from unchecked AI hallucinations. The practice note reflects a broader judicial recognition that technology is a tool, not a substitute for rigorous legal reasoning and ethical vigilance. As AI continues to evolve, the Federal Court’s framework may serve as a model for other jurisdictions grappling with similar challenges, underscoring that the responsible use of AI is essential to maintaining public confidence in the legal system.
https://www.theguardian.com/law/2026/apr/16/australia-federal-court-warning-lawyers-ai-artificial-intelligence