Experts who use artificial intelligence to help prepare their reports should be aware of a recent Family Court ruling which shows where the courts draw the line between acceptable assistance and AI authorship.
In Cornwall Council v M & Ors [2026] EWFC 306, Her Honour Judge Flexman was unable to place significant weight on an expert’s evidence after finding that AI had been used to generate substantive content.
Fiona Morrison, Director and Co-founder of TLA Medicolegal, says: “Experts remain personally responsible for every word they put their name to whatever tools they use to prepare it.”
The care proceedings involved an independent social worker who had produced a parenting assessment, assessments of potential kinship carers and an addendum report. The expert did not disclose her use of AI at first which only came to light when the judge asked how the reports had been prepared. The expert then explained that Microsoft Copilot had been used within a secure organisational environment to arrange information and help with drafting.
The judge recognised that the use of AI sits on a sliding scale from acceptable to unacceptable use and that it is not always easy to tell where AI helping an expert present their reports ends and AI effectively writing it begins. She regarded transcription, spelling and grammar correction as relatively low risk uses that can improve efficiency without affecting the substance of the evidence.
The position changed when the expert accepted that a kinship assessment had been generated by AI from her notes, although she maintained that it reflected her own professional opinion. HHJ Flexman was unconvinced finding that the analysis, structure, subheadings and overall presentation had all been produced by AI and that the tone and style differed from the expert’s own written and oral evidence.
The judge was also concerned that AI drafting notes had been left in the report and that factual inaccuracies meant the court could not tell whether the errors came from the expert’s notes or AI.
The judge said that her decision also reflected wider concerns about the quality of the evidence, which she described as lacking in rigour and clarity of thought. She pointed to the duties experts owe under the Family Procedure Rules, which include explaining their reasoning and confirming that the opinions expressed are their true and complete professional opinions. She also explained that using AI could be akin to third party involvement, depending on how it is used.
Fiona says: “This judgment gives experts the clarity many have been looking for on where the courts draw the line with AI. Using it to transcribe, organise or tidy up a report is one thing but once it begins shaping the analysis the opinion is no longer truly the expert’s own and the court will notice.
“Medical experts are instructed for their judgement and must be able to stand behind every word of their report under cross examination. That means reading the records themselves, checking every fact against the source material and being completely open about how AI has been used.”
The judgment referred to the Academy of Experts’ January 2026 guidance, which treats language correction as lower risk and substantive analysis or content generation as higher risk.
It comes as the Civil Justice Council considers asking experts to explain in their statement of truth any use of AI beyond transcription or administrative tasks.
TLA continues to work closely with its experts to make sure every report reflects their own expertise and stands up to the scrutiny that courts and solicitors rightly expect.