# Self-represented Australians are using AI in court-and it's backfiring

> Since 2022, 84 Australian cases have involved AI, mostly self-represented. Courts report fake citations, delays, costs, and privacy risks-verify sources before filing.

- Type: AI news
- Page: https://completeaitraining.com/news/self-represented-australians-are-using-ai-in-court-and-its/
- Published: 2025-09-29
- Updated: 2025-09-29
- Author: Jeroen Erne
- Jobs: Legal
- Reading time: 2 min

### Australians are using AI in their court cases (and it isn't going well)

Judges get why self-represented parties turn to free tools. As one County Court judge put it this year, generative AI can be "beguiling" when the task feels overwhelming. The catch: if you don't verify what it produces, you risk damaging your case.

#### What the data shows

Since late 2022, at least 84 Australian cases have involved generative AI. More than three-quarters (66 of 84) were self-represented litigants. Matters spanned property and wills, employment, bankruptcy, defamation, and migration.

Self-representation is already common. In 2023-2024, 79% of litigants in migration matters at the Federal Circuit Court were unrepresented. Courts are responding with plain-language forms and guidance, but the rise of free AI tools is adding a new failure point.

#### What courts are saying now

Queensland courts recently warned that inaccurate AI-generated content can delay matters and even attract costs orders. In New South Wales, the Chief Justice noted a party was candid about using AI, but found many AI-generated submissions "misconceived, unhelpful and irrelevant." Good faith won't fix bad inputs.

#### The risks if AI gets it wrong

- **Fabricated authorities:** Fake cases or misquoted holdings lead to rejected submissions.
- **Outcome risk:** Invalid evidence or argument means you can lose a valid claim.
- **Adverse costs:** Courts may order a self-represented party to pay the other side's costs.
- **No safety net:** If a lawyer relies on fake law, negligence is in play. If you are self-represented, you bear the loss.
- **Confidentiality breaches:** Putting private or suppressed information into a chatbot can expose it publicly and breach orders.

#### Lower the risk: do this instead

- **Use primary sources first:** Check legislation and cases directly on reliable databases such as [AustLII](https://www.austlii.edu.au/?ref=completeaitraining.com) and [JADE](https://jade.io/?ref=completeaitraining.com).
- **Follow court guidance:** Many courts (e.g., Supreme Courts of QLD, NSW, VIC) publish rules on acceptable AI use. Read and apply them.
- **Verify every citation:** Search the case yourself. Confirm it exists, the court and year are correct, and the passage supports your proposition.
- **Protect confidentiality:** Do not paste names, facts under suppression, or privileged material into public AI tools.
- **Use libraries:** Court and university law libraries offer free guides and textbooks that summarise principles accurately.
- **Get human help where possible:** Duty lawyers and community legal centres can sanity-check strategy or filings.

#### If you still plan to use AI

- **Limit scope:** Use AI for structure, checklists, or drafting templates-not for legal research or citations.
- **Cross-check everything:** Treat AI outputs as unverified notes. Validate against primary materials before filing.
- **Keep an audit trail:** Record prompts, versions, and your verification steps in case the court asks.
- **Avoid sensitive data:** Redact and generalise facts when testing wording or structure.

#### For legal teams and chambers

- **Adopt an AI-use policy:** Define approved tools, prohibited tasks (e.g., research), and verification standards.
- **Create a citation checklist:** Mandatory case verification, pinpoint references, and quote accuracy checks.
- **Train your people:** Baseline AI literacy plus legal research refreshers reduce errors and rework.
- **Monitor and review:** Periodic file audits to catch hallucinations, confidentiality risks, and drift from policy.

Access to justice is the problem; AI isn't the shortcut. Until tools can guarantee source-grounded accuracy and privacy, treat AI as a drafting assistant at best-and never as your authority.

If your practice is building safe, baseline AI literacy for staff, see curated options by job role: [Complete AI Training](https://completeaitraining.com/courses-by-job/).

---

This is the Markdown form of https://completeaitraining.com/news/self-represented-australians-are-using-ai-in-court-and-its/. The whole site is mapped at https://completeaitraining.com/llms.txt.
