Judge criticises expert’s use of AI

As the courts continue to grapple with the use of AI tools, with the Civil Justice Council proposing procedural rule changes to ensure transparency over AI use, the Family Court in Cornwall Council v M and others had to consider expert evidence from an Independent Social Worker (ISW) who had used Microsoft Copilot to compile her report.

“Full professional responsibility”

The ISW had not initially disclosed the use of AI tools in the preparation of her report, but in an addendum report, following questions from the court, she said:

“Microsoft Copilot was used within a secure organisational environment as an administrative support tool to:

• organise and structure information already obtained
• assist in drafting the report

It is important to clarify that:

• Copilot did not generate evidence, undertake analysis, or form conclusions
• All analysis, interpretation, and professional judgment are entirely my own
• All content was reviewed, amended, and verified by me prior to finalisation
• I retain full professional responsibility for the report”.

However, this assertion of independence and quality control was somewhat undermined by the inclusion in one of her reports of the words “This wording keeps the tone balanced and analytical while clearly highlighting the changed circumstances since the earlier assessment”. This was clearly an AI-generated drafting comment.

The judge concluded that while the use of Copilot to transcribe recorded conversations and check spelling and grammar was unproblematic, the substantive content of the ISW’s report had been generated by AI. This “placed a significant gap between the voice and opinion of the expert and the report which she has filed”. The end result was that the judge felt unable to place any significant weight on the ISW’s evidence.

Guidance and transparency

The judge referred to the Academy of Experts’ Guidance for Expert Witnesses on the Use of AI, created by a working group led by Simmons partner Minesh Tanna, and adopted the categorisation of risk by AI use type that is set out there.
Speaking at a conference for expert witnesses on 06 November 2026, Joanna Smith J, the interim Head of the Business and Property Division of the High Court, encouraged expert witnesses to disclose any use of AI in their work. She pointed to the proposal from the Civil Justice Council that all expert witnesses (and witnesses of fact) be required to make a declaration on their use of AI where it influences the substantive content of their report, but suggested experts should not wait for the new rules to make such declarations. She told the audience “Obviously, there is at the moment no requirement in the rules that that should be done. But if it is not done, you may find yourselves under cross-examination and potentially open to criticism for not having adequately declared your use of AI”. She went on to say that any expert intending to use AI should disclose this to their instructing solicitors in advance and discuss with them the proposed use.

Scrutiny to be expected

The use of AI in producing any court documents comes with risks and needs to be approached with care. For expert witnesses, the risks include false conclusions due to not understanding how the AI tool works, the loss of truly independent judgement and the inclusion of unidentified errors. Any expert appearing in the High Court who has not made a declaration as to their AI use can expect to be asked whether they have used AI and many judges will be unimpressed by a failure to disclose this, even in advance of any rule changes.

The contents of declarations on AI use are also likely to become fertile ground for cross-examination. Going beyond what the proposed new rule requires, to declare where AI has been used for spelling and grammar checking for example, may be wise. Experts should familiarise themselves with the Academy of Experts’ Guidance and be prepared to justify their methodology, demonstrating how they have ensured sufficient human oversight of any AI outputs. Where the court is not confident in an expert’s approach to AI, it is likely that they will accord little if any weight to that expert’s evidence.