Generative AI is transforming how witness testimony is collected in legal proceedings, though its adoption introduces risks that courts and legal professionals are still learning to address. The transition from relying on human memory to documentary evidence in disputes began more than ten years ago, when judges such as Lord Leggatt highlighted the unreliability of witness recollections. Today, tools like ChatGPT are being tested in courtrooms, prompting debate over whether they enhance or compromise the credibility of evidence.
In 2021, Practice Direction 57AC was introduced in the Business and Property Courts, which was intended to address these kinds of concerns and to enhance the relevance and reliability of witness evidence through additional rules on the preparation of witness statements. The directive reflected mounting doubts about the accuracy of human memory, particularly in commercial litigation where judges were encouraged to place little if any reliance on witnesses’ recollections and to base factual findings on inferences drawn from the documentary evidence and known or probable facts. This approach now confronts new challenges as AI systems become integrated into legal workflows.
Judges reject AI-drafted witness statements
Judicial responses to AI-assisted evidence remain inconsistent, but recent rulings signal growing caution. In the 2026 case Godwin v Godwin, a High Court judge determined that two witnesses—both characterized as “sophisticated individuals”—had inappropriately used ChatGPT to draft their statements. The judge rejected the claim that Microsoft Word’s standard tools were insufficient, concluding that the AI-generated statements might not accurately reflect the witnesses’ own words, as required by the Civil Procedure Rules. As a result, the judge treated their testimony with heightened scrutiny.
The probate dispute carries wider implications for legal practice. The judge’s concerns extended beyond mechanical errors to whether AI could subtly alter the substance of witness statements without proper disclosure. This aligns with broader discussions about transparency in legal documents, where even limited AI assistance might undermine evidentiary authenticity.
Legal bodies weigh AI’s impact on evidence
The Civil Justice Council (CJC) has taken up the issue, conducting a 2025 consultation that found no need for new AI-specific rules regarding pleadings, formal legal submissions. However, witness statements remain a focal point. The CJC’s June 2026 update emphasized concerns “about the potential for AI tools to reshape, embellish or otherwise influence evidence in ways that may not be immediately apparent”; but also noted that such issues are not novel issues created by AI. Among respondents raising concerns was the Law Society, which called for all witness statements to contain a declaration as to non-AI use. The CJC’s full report is expected at the end of the year.
Underlying the debate is the recognition that AI does not introduce entirely novel risks but rather intensifies existing ones. Courts have long grappled with the fallibility of human memory, and AI-assisted drafting could exacerbate those challenges. While the technology may improve clarity or efficiency in document preparation, its unchecked use risks producing evidence that appears polished but lacks reliability. This dilemma explains why judicial bodies are advancing cautiously, focusing not only on whether AI is employed but on how it is deployed, and whether legal teams can verify the authenticity of resulting statements.
For solicitors and witnesses, the current guidance is straightforward: exercise caution. Both the CJC’s ongoing review and the precedent set by Godwin v Godwin demonstrate that courts are closely monitoring AI’s role in evidence gathering. The tension between operational efficiency and evidentiary integrity remains unresolved, with significant consequences for legal proceedings.
