Employment Appeal Tribunal Issues Stern Warning on AI Use and Procedural Compliance
The Employment Appeal Tribunal has set new expectations for litigants using AI and stressed the mandatory nature of procedural compliance in all legal submissions.
• public
Tribunal Addresses AI and Procedural Conduct
In a recent judgment, the Employment Appeal Tribunal (EAT) has provided critical guidance regarding the use of generative artificial intelligence and the importance of adhering to procedural rules. The case of Hancox v Sutherland & Others [2026] EAT 139 serves as a stark reminder that all litigants, whether represented or acting in person, bear personal responsibility for the accuracy and compliance of documents submitted to the court.
The Risks of Unchecked Generative AI
The appeal proceedings were complicated by the submission of a 300-page skeleton argument created using ChatGPT. The Deputy Judge of the High Court, Marcus Pilgerstorfer KC, noted that the document was entirely unsuitable. It failed to comply with practice directions, was excessively long, and lacked focus. The judgment emphasises that while AI tools can assist those without legal representation, they are not authoritative sources of legal knowledge.
The EAT highlighted several core responsibilities for anyone using AI to prepare legal documents:
- Documents must comply with all procedural rules and practice directions.
- Litigants must check the output thoroughly for factual, evidential, and legal accuracy.
- Parties must ensure that the court is not misled by fictitious citations or incorrect legal propositions.
- Submissions must remain concise and focused on the central arguments.
Procedural Rigour and Future Sanctions
The EAT reiterated that being a litigant in person does not exempt one from the obligation to follow the EAT Rules and Practice Direction. The judgment makes clear that attempts to shift the burden of verifying documents onto the opposing party are unacceptable. Future submissions that fail to meet these standards may be rejected, and the Tribunal warned that persistent non-compliance or excessive, repetitive correspondence could lead to further sanctions, including restricted communication methods.
While the EAT allowed one ground of appeal to proceed to a full hearing, it dismissed the appellant's application to introduce additional grounds and certified his application for disclosure as totally without merit. This case stands as a significant marker for the future of legal practice in an era where AI tools are increasingly accessible but require disciplined, responsible application.
Read the entire judgment here: Hancox v Sutherland & Others [2026] EAT 139
