All witnesses should be firmly discouraged from using AI to prepare for giving evidence. AI coaching is still coaching, and coaching is not permitted in proceedings before the English courts. That point was made by the Court of Appeal (Criminal Division) in R v FGD [2026] EWCA Crim 918, but this judgment reaches well beyond criminal practice. Because witness training is forbidden in all jurisdictions, civil litigators should pay close attention to it too.
Summary
The defendant had been charged with rape. During cross-examination, the complainant produced screenshots to address a challenge about where they had been at a particular time on the night in question. That led to an order that the complainant’s phone be examined. This examination revealed two documents recording the complainant’s recollection of the relevant events, followed by a series of cross-examination-style questions and suggested answers. The appeal proceeded on the basis that those documents had been generated by AI in response to prompts from the complainant. Following an application by defence counsel, the Recorder stayed the proceedings as an abuse of process, holding that the complainant had been coached and that a fair trial was impossible.
The Court of Appeal agreed that the AI documents amounted to witness coaching but held that the Recorder had erred in treating that conclusion as the end of the matter. May LJ, with whom the other Lords Justices agreed, found that the stay jurisdiction was exceptional. Amongst other things, May LJ considered that the complainant’s evidence-in-chief had been given by pre-recorded interview years before trial and long before the use of AI, and that there were contemporaneous texts to friends. On that basis, any contamination risk principally affected the complainant’s answers in cross-examination, rather than the whole of their account and the question of the complainant’s credibility could have been dealt with by the jury, who would have had at their disposal the transcript of the complainant’s interactions with the AI agent.
Momodou
The leading authority cited by May LJ was R v Momodou and Limani [2005] EWCA Crim 177. In that judgment, Judge LJ drew the distinction between prohibited witness training or coaching and permissible witness familiarisation, and explained why coaching is objectionable:
The witness should give his or her own evidence, so far as practicable uninfluenced by what anyone else has said, whether in formal discussions or informal conversations. The rule reduces, indeed hopefully avoids any possibility, that one witness may tailor his evidence in the light of what anyone else said, and equally, avoids any unfounded perception that he may have done so. These risks are inherent in witness training. Even if the training takes place one-to-one with someone completely remote from the facts of the case itself, the witness may come, even unconsciously, to appreciate which aspects of his evidence are perhaps not quite consistent with what others are saying, or indeed not quite what is required of him. An honest witness may alter the emphasis of his evidence to accommodate what he thinks may be a different, more accurate, or simply better remembered perception of events. A dishonest witness will very rapidly calculate how his testimony may be “improved”. These dangers are present in one-to-one witness training. Where however the witness is jointly trained with other witnesses to the same events, the dangers dramatically increase. Recollections change. Memories are contaminated. Witnesses may bring their respective accounts into what they believe to be better alignment with others. They may be encouraged to do so, consciously or unconsciously. They may collude deliberately. They may be inadvertently contaminated. Whether deliberately or inadvertently, the evidence may no longer be their own. Although none of this is inevitable, the risk that training or coaching may adversely affect the accuracy of the evidence of the individual witness is constant.
Be warned
The Court of Appeal warned that witnesses who use AI to prepare for evidence may create serious forensic and practical problems. In an extreme case, the court may have to consider excluding evidence. AI use may also open up witnesses to further lines of enquiry, exposing a witness’s online activity to wider and more intrusive investigation than would otherwise have been justified. Although this warning was given in a criminal appeal, the practical lesson is plainly relevant to civil litigation: solicitors should tell witnesses not to use AI to rehearse, refine or test their evidence, and should record that warning in clear terms.
The takeaway
The takeaway for English practitioners is simple: do not assume that AI is a harmless private rehearsal space. If a witness feeds their account into an AI tool and asks it to generate cross-examination questions and model answers, that raises the same risks as human coaching. The risk is especially acute in international litigation, where clients, lawyers or witnesses may be more familiar with systems in which witness preparation is the norm. FGD is therefore a useful prompt to update witness-warning protocols, engagement letters and evidence-preparation checklists so that AI use is addressed expressly.
