AI Transparency in Courts
The problem: Courts increasingly use artificial intelligence, but don't disclose it. Self-represented litigants have no way to know if algorithms influenced their case.
Learn more
What's Happening
The NSW Supreme Court issued Practice Note SC Gen 23 (effective February 2025) requiring lawyers to disclose use of generative AI (like ChatGPT) in submissions. However, this covers only lawyer use — not judicial use of AI tools or algorithmic decision-support systems. Other state courts have not yet published equivalent policies, and no court has published standards for disclosure of AI use in judicial decision-making.
Why This Matters
- Undisclosed algorithmic influence: A judge may use case-prediction software without disclosing it in reasons for judgment. Litigants have no way to challenge algorithmic bias or errors.
- Self-represented litigant vulnerability: Unrepresented litigants cannot afford AI research tools and don't know when opposing counsel or the court uses them, widening information asymmetry.
- Reliability risks: AI-generated legal citations can be fabricated. A barrister was referred to disciplinary regulators in 2024 for submitting non-existent case law generated by ChatGPT.
- No appeal mechanism: If a litigant discovers AI was used but wasn't disclosed, there is no clear legal avenue to challenge the decision.
Reform Ask
- National AI Disclosure Standards: Parliament should commission the ALRC to develop national court AI transparency standards, including mandatory disclosure in all submissions and published judicial policies on algorithmic tools.
- Self-Represented Protections: Courts should provide plain-language explanations of when AI may have been used in case assessment, and access to affordable AI-powered legal research tools.
- Audit & Appeal Mechanisms: Establish court-level AI audit functions and statutory grounds for appeal where undisclosed AI use is discovered.