Summary
The Federal Court of Australia turned general AI caution into operational litigation rules on April 16, 2026. Its new GPN-AI practice note makes GenAI use permissible in principle, but only with clear responsibility for accuracy, confidentiality, disclosure, and evidentiary integrity.
Why It Matters
For lawyers, this is a direct and highly usable record of AI being operationalized inside court procedure:
- the court expressly allows GenAI use only when lawyers and parties remain responsible for the result
- it warns against feeding confidential or privileged material into publicly available systems without proper controls
- it requires verification of citations, authorities, quotations, evidence characterizations, and factual summaries
- it creates specific disclosure expectations when GenAI is used in ways that may matter to the court or to opposing parties
- it treats AI-generated or AI-manipulated evidentiary material as something that may require explanation, provenance, or challenge
This is not just another sanctions story. It is a current example of a court converting AI concerns into usable procedural expectations.
What the Source Says
The practice note says GenAI can assist with legal work, but responsibility for all filed material remains with the person who signs or submits it. It separately addresses confidentiality and privilege, accuracy and verification, and disclosure. It also flags evidentiary issues by warning that AI-generated or AI-altered content may need to be identified and that parties should be ready to explain how such material was created or used.