Make AI use visible before expert evidence reaches court

Legal Futures reports that the Civil Justice Council has recommended that the use of AI in preparing witness statements and expert reports should be declared to the court. The recommendation covers witnesses of fact and experts. The Law Society Gazette, top stories later reports that a Business and Property Division interim deputy urged expert witnesses to start telling courts about AI use now, before new rules mandate transparency. The immediate development is therefore a proposed disclosure requirement joined by a judicial warning not to wait for it.

That matters because AI-assisted legal work does not enter litigation on production cost alone. The court decides what must be disclosed, and professional enforcement decides what transparency means in practice. A declaration requirement makes the production process part of the evidence-facing procedure rather than an invisible matter behind the finished statement or report. The recommendation does not establish that disclosure will improve accuracy, reduce disputes or lower the cost of litigation. It does establish that the institutional setting is becoming part of the route by which AI-assisted work reaches the court.

The Civil Justice Council recommendation proves that declaration of AI use has been proposed for witness statements and expert reports. The Law Society Gazette report proves that an interim deputy urged experts to disclose use before a transparency mandate takes effect. Together, they show a movement from a recommended practice toward a rule described as forthcoming. They do not prove that a mandatory rule is already in force, that every court will apply the same procedure, or that disclosure alone will identify unreliable work. They also do not establish that AI use caused any error in a witness statement or expert report. For the stronger reading, a published rule would need to specify who must declare use, what use must be described, and how a court responds when a declaration is absent or incomplete. Those details are not supplied here.

The distinction matters for AI-native legal services. A firm can deliver a finished statement or report, but the court may still require visibility into how that work was prepared. That requirement places a procedural condition alongside the delivered work: the service must be capable not only of producing the document but also of supporting the declaration that the court requires. The recommendation and warning do not show whether any provider can meet that condition, whether clients will receive a useful account of AI use, or whether the process will add material delay or cost. They show only that court-facing transparency is being treated as a rulemaking and compliance question.

The next confirming fact is a published rule or court direction that turns the proposed transparency mandate into a defined obligation. It should state the scope of AI use covered, the required form of declaration and the consequence of non-disclosure. A case in which a court requires a witness or expert to explain AI use would provide a narrower practical test. If the new rule leaves disclosure undefined or produces no visible response to non-disclosure, the stronger claim that procedure materially governs AI-assisted litigation will remain untested. We will treat the rule’s operative requirements, not the warning alone, as the measure.