A claimant was fed answers through smart glasses while being cross-examined, and his evidence was rejected in full. That is the part everyone has written about. In the same judgment, a jointly instructed expert's opinion was also rejected, for a reason that has nothing to do with technology and everything to do with how the expert was instructed.
A High Court judge was given a quotation from a rule that does not exist. The chat transcripts show the AI warning, more than once, that the wording had not been verified against a primary source and should be checked before anyone relied on it. Nobody checked. The judgment is a lesson about supervision rather than about technology, and it reaches anyone who relies on work done by someone else.
A complainant used an AI tool to rehearse answers before cross-examination, and the trial was stopped. The Court of Appeal reversed that decision and ordered a retrial. The case is being reported as a ruling that AI preparation is witness coaching. Read against the judgment, that is not quite what happened, and the difference matters for anyone giving or handling evidence.
Yes. An expert witness can use AI for the reading: extracting facts, building a chronology and finding entries across a medico-legal bundle. The opinion must remain the expert's own, every AI-produced fact must be verified against its source page, and the Part 35 duty to the court is untouched by the tooling.
How to write an expert witness report comes down to a structure the court can follow and check. Lead with the opinion, work through the material and the reasoning, state the limitations plainly, and close with the statements Part 35 requires. This guide gives you the section-by-section shape, a template you can adapt, and the mandatory wording.
The courts are drawing attention to how AI is used in expert and witness evidence: the Court of Appeal in R v FGD, the CJC's review and the BSB's guidance all land on the same idea, that AI use should be transparent to a proportionate degree. For an expert witness that means being able to say what a tool did, what the expert checked, and what stayed the expert's own judgement.
With a record number of claims now settled out of court, early case resolution requires clear evidence and expert opinion. Bundles can run into thousands of pages, but Alldoq Verify streamlines the workflow.
The bundle is almost never complete. A referral with no clinic letter, a test with no result, three weeks of an admission with no notes: gaps like these are routine, and how an expert handles them decides whether the opinion holds. Name the gap and reason around it openly and the report is stronger for it. Work around it in silence and it becomes the thread the other side pulls.
The joint statement is where two experts agree what they agree on and pin down exactly where they part. Walk into the discussion with the records in order and your reasoning already on the page, and the meeting sharpens the case. Walk in unprepared and it is where a sound opinion quietly comes apart.
AI medical record review uses language models to read a medico-legal bundle, GP records, hospital notes, imaging and correspondence, and return an indexed chronology, typed fact tables and summaries with a source page attached to every fact. Used well, it turns days of reading into hours. Used badly, it produces confident answers no one can check. The line between the two is the source page.
Under Part 35 of the Civil Procedure Rules, an expert's overriding duty is to the court, not to the party paying them. That duty shapes the independence of the opinion, the mandatory statements in the report, and the obligation to flag where the evidence is incomplete or your view has a range.
ALLDOQ is a sponsor of Ireland's Expert Witness Conference 2026, held on 20 May at the Radisson Blu Royal in Dublin. The day brings judges, barristers, solicitors and expert witnesses together to examine the standards behind credible expert evidence, from methodology and independence to AI and cross-examination.
