The Federal Court of Australia’s new Generative AI Practice Note sets clear rules for using ChatGPT, Copilot and AI in litigation – essential reading for class action lawyers and corporate clients.

The Federal Court of Australia’s new General Practice Note on the Use of Generative Artificial Intelligence (GPN‑AI) released on 16 April 2026 formalises the Court’s expectations around the responsible use of tools such as ChatGPT, Copilot and other generative AI systems in Federal Court proceedings.

With approximately 70% of all Australian class actions commenced in the Federal Court, GPN-AI is essential reading for class action practitioners and corporate clients confronting group proceedings, setting clear expectations for how AI can, and cannot, be used in class action litigation.

While GPN-AI recognises that AI can improve efficiency, reduce costs and enhance access to justice, it makes clear that its use must not compromise the proper administration of justice.

At its core, GPN‑AI reinforces that:

• Existing professional and legal obligations apply with equal force where AI is used.
• Practitioners must have a basic understanding of the capabilities, limitations and risks of generative AI, including its tendency to produce inaccurate or fictitious material.
• The presentation of false or misleading information to the Court – whether generated by a human or an AI tool – is unacceptable.

Non-compliance with GPN-AI (or the Court's orders or directions) may attract serious consequences, including adverse costs orders and professional repercussions for practitioners.

Who must comply with the Federal Court's AI practice note?

GPN‑AI applies broadly to all persons who appear before the Court or file documents with it. This includes:

• legally represented parties
• litigants in person
• witnesses, including experts, and
• third parties subject to subpoenas or document production orders.

While GPN-AI is (expressly) not an exhaustive statement from the Federal Court on all things AI – it does contain prescriptive rules with respect to the following types of documents and circumstances in which generative AI may be used by parties before the Court. 

Pleadings and submissions: GPN-AI emphasises that particular caution is required when AI is used in preparing pleadings, submissions, chronologies and lists of documents. If AI has been used, the responsible lawyer must personally verify that facts are properly provable, authorities exist (and support the propositions advanced), evidence is admissible, and discovery obligations under the Federal Court Rules are met. The practice note leaves no room for “outsourcing” professional judgment to an AI system.

Evidence and expert reports: More stringent requirements apply to evidence (including expert reports). GPN-AI expressly acknowledges that AI must not undermine the integrity of affidavits, witness statements or expert reports. Where AI tools have been used to summarise, analyse or generate material that informs evidence or opinion (or its admissibility), GPN-AI mandates that any such uses must be transparently disclosed at the start of the document.  The disclosure must state (as concisely as possible) where in the document AI has been used, and how. Experts otherwise remain subject to their overriding duty to the Court and GPN-AI emphasises that experts must ensure that their reports reflect their own reasoning and opinions.

Sharing materials with AI tools: The practice note also highlights risks in entering confidential, privileged or suppressed material into AI tools, even where those tools are described as secure or ring-fenced. Doing so may breach court orders, implied undertakings or confidentiality obligations, even where disclosure is unintended. Practically, password protection of confidential, privileged or supressed materials can assist to prevent inadvertent disclosure via AI tools.

Key takeaways for practitioners and in-house teams

• For experienced litigators, GPN‑AI should come as no surprise.  It seeks to reaffirm the well-known fundamentals of legal practice before the Courts: accuracy, candour, independence of judgment and responsibility for every document put before the Court.
• For corporate clients and in-house legal teams, GPN-AI serves as an important reminder of the Federal Court's expectations.  With the 

Pleadings and submissions: GPN-AI emphasises that particular caution is required when AI is used in preparing pleadings, submissions, chronologies and lists of documents. If AI has been used, the responsible lawyer must personally verify that facts are properly provable, authorities exist (and support the propositions advanced), evidence is admissible, and discovery obligations under the Federal Court Rules are met. The practice note leaves no room for “outsourcing” professional judgment to an AI system.

Evidence and expert reports: More stringent requirements apply to evidence (including expert reports). GPN-AI expressly acknowledges that AI must not undermine the integrity of affidavits, witness statements or expert reports. Where AI tools have been used to summarise, analyse or generate material that informs evidence or opinion (or its admissibility), GPN-AI mandates that any such uses must be transparently disclosed at the start of the document.  The disclosure must state (as concisely as possible) where in the document AI has been used, and how. Experts otherwise remain subject to their overriding duty to the Court and GPN-AI emphasises that experts must ensure that their reports reflect their own reasoning and opinions.

Sharing materials with AI tools: The practice note also highlights risks in entering confidential, privileged or suppressed material into AI tools, even where those tools are described as secure or ring-fenced. Doing so may breach court orders, implied undertakings or confidentiality obligations, even where disclosure is unintended. Practically, password protection of confidential, privileged or supressed materials can assist to prevent inadvertent disclosure via AI tools.

Key takeaways for practitioners and in-house teams

• For experienced litigators, GPN‑AI should come as no surprise.  It seeks to reaffirm the well-known fundamentals of legal practice before the Courts: accuracy, candour, independence of judgment and responsibility for every document put before the Court.
• For corporate clients and in-house legal teams, GPN-AI serves as an important reminder of the Federal Court's expectations.  With the increasing deployment of generative AI within internal legal, compliance and document-review functions, individuals should be mindful of how generative AI is being used, particularly where outputs may ultimately be relied on in litigation or other regulatory proceedings.