Court and Tribunal Guidance on AI Use in Proceedings

Lesson concept diagram

Courts and tribunals have begun to issue formal guidance on AI use in proceedings. This guidance is not yet uniform across jurisdictions but the themes are consistent. Understanding these expectations is essential because failure to follow them can damage your case and your professional standing.

The English courts expect disclosure of AI use. If an AI tool assisted in drafting a submission, a skeleton argument or evidence in chief, the court should be told. This is not a formal requirement written into Civil Procedure Rules, but judges have made clear in unreported cases that they view non-disclosure as concerning. A judge confronted with well-drafted submissions that turn out to be AI-assisted may wonder whether the lawyer understands their own arguments. Disclosure upfront eliminates this risk. A simple statement such as “This skeleton argument was drafted with assistance from an AI language model and verified by the author before filing” is sufficient.

The courts are particularly alert to AI use in evidence. A witness statement that purports to be the witness’s own account but was drafted by an AI raises evidentiary issues. The statement must reflect the witness’s own knowledge and recollection, not the language model’s understanding of what is typical in such statements. If the statement is drafted by AI and not properly reviewed with the witness, the evidence may be unreliable. Some judges have indicated in guidance that they will order that AI-drafted statements be verified in cross-examination, adding cost and delay.

AI-generated citations in court submissions are treated seriously. If a citation is wrong and the judge discovers it, the case outcome may not change but your professional standing will be damaged. Judges assume that lawyers have verified their authorities. If you cite a fabricated case, the judge may report you to the Law Society. Courts have indicated that they expect lawyers to be able to explain the basis for each citation if asked during oral argument. If the citation came from an AI tool, you should be able to say so and to confirm that you verified it.

The Civil Procedure Rules require lawyers to conduct a proportionate investigation. This includes understanding the state of the evidence and the law. The rules do not explicitly address AI but the principle applies. You cannot delegate your duty to understand the case to an AI tool. You must be able to explain your strategy, your legal analysis and your assessment of the evidence to the court. Using AI as a shortcut to avoid this work is a breach of the proportionality duty.

Some tribunals have begun to ask counsel explicitly whether AI was used in preparation. Employment Tribunals, particularly those dealing with complex discrimination claims, have started to question whether AI-assisted evidence is reliable. The Tribunal wants to understand whether the evidence is the witness’s own account or a language-model-generated version of what such an account typically looks like. This is a fair concern.

The Court of Appeal has not yet issued detailed guidance on AI use, but the principles from lower courts suggest that appellate courts will be equally concerned with disclosure and verification. An appeal that relies on citations later found to be fabricated will face serious problems.

Internationally, courts in the United States, Canada and Australia have issued warnings about AI-generated citations. Several high-profile cases have resulted in sanctions against lawyers who relied on AI citations without verification. These sanctions typically include costs orders and occasional referrals to professional bodies. The pattern is clear: courts expect verification.

A practical protocol is to disclose AI use upfront in any document you file. Do not wait to be asked. Anticipate the question. If a significant portion of your legal research was assisted by AI, say so and confirm that you verified all authorities. If evidence was drafted with AI assistance, confirm that it reflects the witness’s own knowledge and that the witness has approved it. If a skeleton argument was AI-assisted, confirm that you have reviewed it for accuracy and verified all legal authorities. This transparency builds confidence.