How often do you see an expert witness themself become part of the reason a criminal conviction is overturned?

That is, in effect, what happened in The King v ZX [2025] NICA 76, a decision of the Court of Appeal in Northern Ireland concerning convictions for murder and attempted murder.

Psychiatric evidence was central to the trial. The defendant’s mental state was directly relevant to the issues the jury had to decide, and both the prosecution and defence relied on psychiatric experts.

What makes the case particularly striking is not simply that the Court of Appeal disagreed with an expert’s opinion. Courts do that all the time. The problem went much deeper. The court concluded that the prosecution psychiatrist’s evidence had evolved in such a way that it could no longer be regarded as the independent product of an expert uninfluenced by the demands of the litigation.

The expert had, in the court’s words, effectively assumed an impermissible role in promoting the prosecution case.

That is an unusually serious finding against any expert witness.

An opinion which developed during the trial

Before trial, the prosecution psychiatrist had assessed the defendant as suffering from depression, although not to the degree suggested by the defence expert. During his oral evidence, however, he introduced a much more significant proposition: that the most likely explanation for what had happened was “spousal revenge”.

That theory had not appeared in his reports or his contemporaneous notes. It emerged in evidence after the expert had been provided with research material addressing that subject by prosecution counsel.

There were other examples of important matters appearing for the first time at trial. The expert placed weight on the condition of the defendant’s home as an indicator against severe depression, although that had not featured in his reports or notes. He also relied on the severity of her self-inflicted injuries in a way which had not previously been identified.

Experts are, of course, allowed to change their opinions. New information may emerge or further consideration may lead to a different conclusion. But a material change of view should be identified and communicated. A party should not discover an important new expert opinion for the first time when the expert is giving evidence.

That principle was central to the Court of Appeal’s reasoning.

A quotation which changed meaning

Perhaps the most uncomfortable factual detail concerned a quotation attributed to the defendant.

The expert’s report recorded her as saying:

“he destroyed my life, so I destroyed his”

When the expert’s original handwritten notes were eventually disclosed during the trial, however, they recorded only:

“he destroyed my life”

There was also a discrepancy about when the statement was said to have been made.

The difference is obvious. Adding the second half of the sentence changes a statement about what had happened to the defendant into something which appears retaliatory. That was particularly significant given the later emergence of the “spousal revenge” theory.

The court recorded that, when challenged in cross-examination, the expert at one stage agreed that he had lied about what had been said, although he later resiled from that position.

Whatever the explanation, it illustrates why an expert’s original notes can matter just as much as the final report.

Material which had not been considered

The problems were not confined to the development of new theories.

The expert accepted that he had not considered police body-worn footage, had not asked to see it and had not regarded it as part of his assessment. He had also failed accurately to record when, and to what extent, the defendant had received antidepressant medication. When these omissions were put to him, he maintained that they did not alter his opinion.

That poses a familiar problem for experts. A conclusion may ultimately remain the same after additional material is considered, but that is different from never having considered the material in the first place.

An expert needs to be able to explain not only the conclusion reached but the evidential basis on which it was reached, including material which points in the other direction.

The relationship between the expert and the legal team

The Court of Appeal was also highly critical of the way in which the prosecution expert had been handled.

Before giving evidence, the expert had a telephone conversation with prosecution counsel lasting between approximately 45 minutes and an hour. Neither participant made a note or record of the conversation. A further meeting took place on the day the expert began giving evidence.

The concern was not simply that discussions had taken place. Conferences between lawyers and experts are normal and often essential. The difficulty was what happened afterwards.

The court noted that matters raised during the consultation subsequently became prominent in the expert’s evidence and increasingly supportive of the developing prosecution case. It said this gave rise to an appearance of coaching or of the expert having been consciously or subconsciously influenced.

Importantly, the court did not make a finding of deliberate coaching, but it did find that the prosecution’s conduct had fallen below recognised standards of good practice.

There is a useful distinction here. Lawyers are entitled to test an expert’s opinion, challenge assumptions, ask questions and introduce new material. What they cannot properly do is shape the expert’s evidence into the case they want to present.

The expert has the corresponding responsibility to resist that process.

No written letter of instruction

Remarkably, the prosecution psychiatrist had no written letter of instruction.

The Public Prosecution Service explained that he had been given a brief verbal explanation of the case, but nothing had been put in writing.

The Court of Appeal’s response was blunt. It described that as “patently not good enough”, said that the practice lacked transparency and stated that in future experts in criminal cases should have a written letter of instruction.

For experts used to civil litigation, that may sound like an obvious requirement. But the reason for it goes beyond administration.

Written instructions establish what the expert was originally asked to consider. They help define the issues, provide a record of the material supplied and make it possible later to see how and why an opinion developed.

Without that starting point, it becomes much harder to distinguish independent analysis from a case theory which may have developed through conversations with the instructing team.

The court’s conclusion on independence

The Court of Appeal returned to well-established principles governing expert witnesses.

Expert evidence should be the independent product of the expert. An expert should provide objective and unbiased assistance to the court, should identify the facts and assumptions supporting an opinion, should not omit material facts which detract from it and should communicate any material change of view promptly.

Applying those principles, the court concluded that the prosecution psychiatrist had departed significantly from his role as an independent expert and had effectively assumed an impermissible role promoting the prosecution case.

The consequences were serious.

The court said that without the perseverance of defence counsel in obtaining the expert’s notes and examining his dealings with prosecution counsel, there was a real possibility that the deficiencies and partiality in the evidence would never have been exposed. It said that this could have led to a miscarriage of justice.

The ground of appeal relating to the expert evidence succeeded.

The responsibility did not stop with the expert

It would nevertheless be wrong to treat ZX simply as a case about an expert who lost his independence.

The Court of Appeal also found serious problems in the conduct of the prosecution and in the way the trial judge dealt with the difficulties once they had emerged.

The prosecution’s handling of the expert was criticised, including the informal nature of the instructions, delayed disclosure and the interaction between counsel and the expert. The ground of appeal relating to prosecutorial misconduct also succeeded.

The court also found that the trial judge had failed to give sufficient consideration to applications to discharge the jury. It concluded that the jury could easily have been misled by the expert evidence and that directions could not have cured the resulting prejudice.

Both convictions were quashed.

The court’s overall conclusion was that the serious errors began with the prosecution’s misconduct and were compounded by the expert witness and the judge.

Could a new expert start again?

There is one final aspect of the case which is particularly interesting for anyone involved in expert evidence.

Once the convictions had been quashed, the court had to consider whether a retrial could take place. One difficulty was that the “spousal revenge” theory had by then become embedded in the case papers.

The Court of Appeal was concerned that any new expert might themselves be influenced by that history simply by reading the existing material. It described the risk of the replacement expert becoming “contaminated”.

When a retrial was eventually ordered, the court directed that the new expert’s letter of instruction should, so far as possible, be prepared on a “blank page basis”.

That is perhaps one of the most thought-provoking parts of the judgment.

Expert independence is not only about resisting pressure from the party instructing you. It also means being alert to the way in which the material provided to you can frame the problem before you have even begun your own analysis.

Independence can be lost gradually

Most experts know that their duty is to the court rather than to the person paying their fees. It is one of the first principles of expert witness work.

What ZX demonstrates rather starkly is how independence can be eroded without there necessarily being one obvious moment when the line is crossed.

A discussion takes place. A new theory is raised. A piece of research is introduced. Something which seemed insignificant in the report becomes central in oral evidence. Material which does not fit the emerging theory receives less attention. Gradually, the expert stops simply assisting the court and begins to move with the case being advanced by one side.

That is why the disciplines around expert evidence matter: clear written instructions, complete records, contemporaneous notes, proper disclosure, transparency about changes of opinion and a willingness to consider evidence which does not support the preferred conclusion.

None of those things is glamorous. They are, however, part of what keeps an expert independent.

In ZX, their absence contributed to something very unusual indeed: the expert’s own conduct became part of the reason two criminal convictions could not safely stand.

Case: The King v ZX [2025] NICA 76, Court of Appeal in Northern Ireland.

Read the judgment

The Academy of Experts and HKA have published An Expert’s Perspective, a major new research paper examining how expert evidence is commissioned, prepared and presented in litigation, arbitration and other dispute resolution proceedings.

The research draws on responses from 456 experts and expert assistants across a range of jurisdictions and professional disciplines. It looks at the expert process from appointment through to the presentation of evidence, including instructions, disclosure, report preparation, meetings of experts, joint statements, witness conferencing, expert assistants and the emerging use of artificial intelligence.

Expert witnesses are regularly the subject of judicial comment, professional guidance and academic discussion. Much less attention has been given to the experience of experts themselves and to the practical environment in which their evidence has to be produced.

An Expert’s Perspective seeks to help fill that gap.

Download the Research Paper

The perspective of the expert

Introducing the report, The Rt Hon The Lord Neuberger of Abbotsbury, President of The Academy of Experts and former President of the Supreme Court of the United Kingdom, says:

“I enthusiastically welcome this paper which addresses a subject of considerable importance to the administration of justice: the role, conduct and practical experience of expert witnesses.”

Lord Neuberger is clear that looking at the process from the expert’s perspective does not mean excusing poor expert practice or reducing the responsibility placed upon an expert witness.

Experts must remain independent. Their overriding duty is to the court or tribunal. They must stay within their expertise, distinguish fact from opinion, explain their reasoning and resist any attempt to turn the expert role into that of an advocate.

But the research also demonstrates that the quality of expert evidence is not determined by the expert alone.

Experts operate within a process. How they are selected, appointed and instructed; what documents they receive and when; the timetable within which they are required to work; the way meetings with opposing experts are conducted; and the procedures adopted for presenting evidence can all affect the quality and usefulness of the evidence eventually received by the decision-maker.

As Lord Neuberger observes, the report provides:

“an essential perspective which has too often been absent from the debate: the perspective of experts themselves.”

What did the research examine?

The survey looked across the life of an expert appointment rather than concentrating on a single stage.

Among the subjects examined were:

• how experts are identified and appointed;
• the instructions and procedural information they receive;
• the expert’s involvement in identifying relevant documents;
• the timing and completeness of disclosure;
• the time available for investigation and report preparation;
• meetings and discussions between opposing experts;
• the preparation and review of joint statements;
• tribunal presentations and witness conferencing;
• the use and disclosure of expert assistants; and
• the use of artificial intelligence in preparing expert evidence.

The respondents included highly experienced practitioners. The result is a picture not of a single profession, jurisdiction or type of dispute, but of recurring issues encountered by experts working across a broad range of proceedings.

Clear instructions are the foundation

One of the themes to emerge from the research concerns what happens at the very beginning of an appointment.

An expert needs to know the questions they are being asked to address, the relevant procedural framework, the form their evidence is expected to take and the duties which apply to them.

Yet the survey indicates that this information is not always provided consistently.

That matters because clear written instructions are more than an administrative convenience. They establish the framework within which the expert is expected to carry out an independent professional task.

Lord Neuberger describes written instructions, clearly identified questions, applicable procedural rules and the required declaration as the “essential foundation upon which reliable expert evidence is built”.

The practical lesson is straightforward: invest time in getting the appointment and instructions right at the outset.

Experts need the evidence on which to form their opinions

Disclosure is another recurring issue.

Most respondents reported broadly equivalent access to documents, but the research also identified cases in which experts had difficulty obtaining relevant material, received information late or found that an opposing expert had access to documents which had not initially been made available to them.

Late disclosure can have consequences well beyond inconvenience.

An expert may have to revisit work already undertaken, reconsider assumptions or alter an opinion after significant parts of the expert process have already taken place. That can increase cost, create delay and put additional pressure on already compressed timetables.

The research therefore points towards earlier and more complete disclosure, together with greater consideration of the expert’s potential contribution to identifying the technical material required.

Realistic timetables matter

Experts are frequently working within timetables set by others.

Efficient case management is important, but speed and efficiency are not necessarily the same thing. An unrealistically compressed timetable can reduce the opportunity for proper investigation and analysis and may ultimately create further cost and delay.

As Lord Neuberger puts it in his foreword:

“efficient case management must not be confused with procedural compression.”

The report identifies realistic timetables as one of a number of relatively modest procedural changes capable of producing significant benefits.

Protecting the independence of meetings and joint statements

Meetings between experts and the resulting joint statements can be among the most useful parts of the expert process. They allow areas of genuine agreement to be identified and the remaining disagreement to be narrowed and explained.

Their value, however, depends upon the independence of the experts being preserved.

The survey results are generally reassuring about direct attempts to influence expert opinion. Nevertheless, they reveal variation in how joint statements are prepared and in the extent to which lawyers may seek to see, comment upon or propose amendments to drafts.

There may be legitimate reasons for correcting factual inaccuracies or dealing with matters of law or procedure. The danger arises if that involvement begins to influence the substance of the experts’ agreement or disagreement.

A joint statement must remain the work of the experts.

The research therefore points towards greater transparency in the management of expert discussions and joint statements and clear boundaries between the respective roles of lawyer and expert.

Expert assistants and AI: transparency remains key

The way expert evidence is prepared is also changing.

Expert assistants have long played an important role in substantial and technically complex disputes. Artificial intelligence now presents another potential form of assistance.

Neither changes the fundamental position.

The appointed expert remains responsible for the opinion expressed in their name.

The report identifies transparency as an important principle in both areas. Where assistants have made a material contribution, their involvement may need to be disclosed. Similarly, where AI has been used in preparing expert evidence, the expert must understand and verify the material produced and remain personally responsible for the resulting opinion.

The technology may change. The duty of the expert does not.

Practical improvements rather than radical change

One of the striking aspects of the research is that many of the improvements identified do not require wholesale reform of the expert witness system.

Lord Neuberger describes the merit of the research as lying in its “practical orientation”.

Among the steps identified in the report are:

• earlier appointment of experts;
• clearer written instructions;
• common questions for opposing experts where appropriate;
• timely and proportionate disclosure;
• realistic procedural timetables;
• transparent management of joint statements;
• properly structured tribunal presentations;
• appropriately managed witness conferencing; and
• greater transparency surrounding expert assistants and AI.

Individually, many of these appear relatively modest. Collectively, they can help create a process in which experts are better able to provide the independent and properly reasoned assistance which courts and tribunals require.

Responsibility is shared – but independence remains personal

The report does not suggest that responsibility for expert evidence rests elsewhere.

Poor expert practice should be identified and criticised where appropriate. Experts remain personally responsible for complying with their duties and for the opinions they express.

But meaningful improvement also requires attention to the system around the expert.

Lawyers, parties, experts, judges and arbitrators all play a part in creating the environment in which expert evidence is commissioned, prepared, tested and ultimately relied upon.

The central conclusion of the research is therefore a practical one: expert evidence works best when the process supports independence, transparency, proportionality and focused assistance to the decision-maker.

The publication also brings together relevant guidance produced by The Academy of Experts on the role of the expert witness, meetings between experts, expert reports, expert witness CVs and the responsible use of AI.

A companion commentary by Professor Doug Jones AO considers the findings from the perspective of the tribunal and looks at ways in which case management, disclosure and the expert process can be improved.

Concluding his foreword, Lord Neuberger says:

“Accordingly, I strongly commend this paper to judges, arbitrators, lawyers, experts and all those concerned with the proper administration of dispute resolution.”

The Academy is delighted to have worked with HKA on this research. We hope An Expert’s Perspective encourages further discussion across the dispute resolution community and helps identify practical changes which make expert evidence more effective, more transparent and more useful to those who ultimately have to decide disputes.

A 2026 CIArb report on party-appointed expert witnesses in international arbitration suggests that experts are often brought into disputes later than they should be. The findings are about arbitration, but many of the issues will be familiar to those involved in conventional litigation as well.

Party-Appointed Expert Witnesses in International Arbitration: Effectiveness and Best Practice draws on 419 responses from expert witnesses, arbitrators, external counsel and in-house counsel, together with consultation across common law and civil law jurisdictions. One of its central themes is the timing of expert involvement. Some 71% of respondents thought that, ideally, experts should be appointed before the arbitration itself had begun: 42% once the issues between the parties were known or had crystallised and 29% before a claim was made.

The figures become more interesting when the responses of experts are compared with those of the people instructing them. Only 44% of expert witnesses thought they were typically given enough time to carry out their work. External counsel were considerably more optimistic at 66%, arbitrators at 77% and in-house counsel at 80%. Some 60% of the experts also thought they should have an opportunity to comment directly on the procedural timetable when it is being set.

There was a similar difference of view about why experts are instructed late. Among experts, 42% identified a lack of understanding of the expert’s role and 39% identified cost. External counsel saw cost as the principal problem, at 53%, while only 13% thought lack of understanding of the expert’s role was the main reason.

Those figures come from international arbitration, but the underlying question is wider: are we sometimes concentrating too much on the date by which the expert report is needed, rather than the point at which the expert might usefully become involved?

What can an expert add before the report is due?

Bringing an expert in earlier does not necessarily mean commissioning a full report earlier. There is useful work which may need to happen long before anybody starts drafting one.

An expert may be able to identify whether the technical question being posed is actually the right one, what evidence or data will be needed to answer it, whether important material is missing and whether the proposed timetable bears any relation to the work required. In some disputes the expert may also be able to test whether a claim or defence which looks attractive in legal terms works when the underlying technical, scientific or financial assumptions are examined.

The CIArb report identifies early involvement as potentially helping to test the viability of claims, clarify technical issues, narrow areas of disagreement and inform procedural planning. It also raises the practical difficulty of experts being presented with a timetable once it has already been agreed and then being expected to work within it.

This can become particularly important where several disciplines depend upon one another. A quantum expert, for example, may need findings or opinions from an engineering or delay expert before parts of the quantum analysis can sensibly be completed. If those dependencies are only discovered after individual report deadlines have been fixed, somebody is likely either to duplicate work or to run out of time.

The same issue arises in litigation

In England and Wales, CPR Part 35 requires a party seeking permission for expert evidence to identify the field in which the evidence is required and the issues which it will address. Where practicable, the proposed expert must also be identified. Practice Direction 35 separately requires parties to consider with their experts at an early stage whether an experts’ discussion is likely to serve a useful purpose and, if so, when it should take place.

By that stage some fairly important decisions may already be taking shape. What is the expert issue? Is one discipline enough? Will more than one expert be needed? What material will each expert require? Is sequential evidence sensible? Should the experts meet before or after reports?

Our own Guidance on Joint Statements reflects the fact that there is no single correct sequence. In some cases an early meeting between experts can expose differences in methodology, assumptions or source material before substantial costs are incurred. In others, particularly less complex cases, exchanging reports first may lead to a more focused discussion.

The appropriate sequence depends on the dispute and the evidence. What matters is that it is considered rather than allowed to happen by accident.

Different approaches in different jurisdictions

There is no universal model for expert evidence. This is particularly relevant in international arbitration, where participants may come from very different procedural traditions and the arbitration itself may combine elements which would be unfamiliar in the domestic courts of any one jurisdiction.

Australia provides an interesting comparison. The Federal Court expects parties proposing to use expert evidence to consider at an early stage how that evidence should be managed. Its Expert Evidence Practice Note deals with matters including the questions to be addressed by experts, conferences between experts, joint reports and concurrent expert evidence.

Australia is particularly associated with concurrent expert evidence, sometimes known as the “hot tub”, under which experts may give evidence together and address particular issues in turn. That can make expert involvement part of the management and narrowing of the technical dispute rather than simply a sequence of reports followed by separate cross-examinations.

Singapore takes a rather controlled approach to the use of expert evidence. Under the Rules of Court 2021, expert evidence cannot be used without the Court’s approval. The parties must consider whether it will materially contribute to deciding an issue involving scientific, technical or other specialised knowledge and whether the issue could instead be resolved through agreed facts or agreed materials.

The Singapore courts also require the parties, as far as possible, to agree on a common expert, and the issues to be referred for expert evidence and the agreed or assumed facts are themselves subject to close control. The Singapore Judiciary provides a useful summary of the expert evidence provisions.

Hong Kong also requires permission before expert evidence is adduced in civil proceedings. An application is normally made after pleadings have closed and should identify the area of expertise and the issues on which expert evidence is required. Parties are also expected to consider whether a single joint expert would be appropriate. The Judiciary summarises the process in its guidance on preparing for a hearing or trial.

In US federal litigation, the structure is different again. Rule 26 of the Federal Rules of Civil Procedure requires disclosure of testifying experts and, for retained experts, detailed written reports. The timing is normally set by the court; if it is not, the default deadline is at least 90 days before trial, with a shorter period for rebuttal evidence. That is a disclosure deadline rather than a rule about when the expert must first be retained, and experts in substantial cases may of course be involved considerably earlier.

These differences make “instruct the expert earlier” rather too simple as a universal rule. Some systems favour early involvement of party-appointed experts; others put more emphasis on defining the expert issues first, limiting the amount of expert evidence or considering a common or single joint expert.

The point which travels rather better between jurisdictions is that expert evidence should be considered early enough for the procedure to fit the evidence, rather than forcing the evidence into a procedure which has already been fixed.

Earlier involvement does not mean losing independence

There is an obvious concern in bringing an expert into a dispute at an early stage. A party-appointed expert witness is not another member of the legal team and should not become an advocate simply because they have been present while the case is developing.

But there is a difference between asking an expert to construct the client’s case and asking the expert whether the case being contemplated can be supported within their field of expertise.

An expert may quite properly say that a proposed methodology will not work, that the available data is inadequate, that another discipline is required or that an assumption on which the claim is being built is technically unsustainable. Finding that out before pleadings are settled and directions are agreed may be considerably more useful than discovering it three weeks before the expert report is due.

The capacity in which the expert is instructed also matters. An expert retained privately to advise a legal team performs a different function from an expert who is to give independent evidence to a court or tribunal. The distinction between advisory or “shadow” experts, party-appointed expert witnesses and single or common experts deserves separate treatment, particularly because duties, disclosure and privilege can differ between jurisdictions.

Start with the expert evidence, not just the expert report

The CIArb research does not suggest that every expert should be appointed at the beginning of every dispute. The appropriate approach will depend on the dispute, the tribunal or court, the jurisdiction and the type of expert evidence required.

It does suggest that leaving consideration of expert evidence until the case has already been designed can create problems which are difficult to repair later. Timetables may prove unrealistic, necessary evidence may not have been preserved or obtained, different expert disciplines may have been sequenced badly and the questions eventually put to the expert may not be the questions which actually need answering.

That should sound familiar outside arbitration as well.

When a dispute is likely to depend materially on expert evidence, the useful question may not be simply “When do we need the report?”

It may be worth asking rather earlier: “When do we need the expert?”

Further reading: CIArb, Party-Appointed Expert Witnesses in International Arbitration: Effectiveness and Best Practice.

An expert may have sound reasons for favouring a particular answer. But explaining those reasons is only part of the task. Where there are competing opinions, the report also needs to show that they have been considered fairly, including the evidence which does not sit comfortably with the expert’s own conclusion.

JBX v Frimley Health NHS Foundation Trust [2026] EWHC 2294 (KB), decided on 4th September, 2026 by Andrew Kinnier KC, sitting as a Deputy Judge of the High Court, concerned damages following admitted clinical negligence. One issue was future care and accommodation for a profoundly brain-injured claimant.

What the court found

The judge identified two particular concerns about the claimant’s neuropsychiatric evidence:

• The expert favoured home care but had inadequately examined its disadvantages and the advantages of residential care.
• The judge criticised his persistence with a diagnostic opinion outside his expertise.

As a result, the defendant’s evidence was preferred on disputed neuropsychiatric matters (paragraphs 95–97).

Similar problems affected the claimant’s care evidence.

• A letter acknowledged “positives and negatives” but listed only home care’s benefits and residential care’s drawbacks.
• The judge found the overall approach one-sided (paragraphs 113–114).

Both care experts were giving oral evidence for the first time. Despite the defendant’s expert’s marked nervousness, her report assessed competing options with reference to medical evidence. Her evidence was preferred where they disagreed (paragraphs 112–117).

The reporting requirement

In civil proceedings in England and Wales, Practice Direction 35, paragraph 3.2(6), requires an expert’s report to summarise the range of opinion, where one exists, and give reasons for the expert’s own opinion. Paragraph 2.3 requires consideration of material facts which might detract from that opinion, while paragraph 2.2 requires objective, unbiased assistance rather than advocacy.

There is a practical difference between those requirements. An expert might describe several professional approaches yet omit an inconvenient fact. Equally, a report might accurately record the facts but give no meaningful account of why another expert could interpret them differently. Both the evidential material and the competing analysis need attention.

This sits within the overriding duty under CPR 35.3 to assist the court on matters within the expert’s expertise. That duty takes precedence over obligations to the person instructing or paying the expert. A useful way of approaching the report is therefore to ask what the court needs to understand, rather than what makes the instructing party’s position easiest to defend.

Considering an alternative does not mean endorsing it

For example, suppose a building expert recommends replacing a defective component rather than repairing it. The report could explain why replacement would solve the problem, then dismiss repair as inadequate. A more useful comparison would identify what the proposed repair could achieve, the assumptions on which it depends and the evidence that causes the expert to reject it. The disadvantages of replacement, such as disruption or unnecessary intervention, would also need examination where they matter to the issue being addressed.

That exercise might leave the opinion entirely unchanged. It would nevertheless give the reader a much clearer explanation of why it was made. Alternatively, it might reveal that a repair is viable in circumstances the expert had not initially considered, or that further investigation is needed before choosing between the options.

Experts must, therefore, be equally clear about the question they are answering. “I would choose X” and “Y is not reasonable” are different propositions. Explaining a preference does not, without further analysis, establish that another approach falls outside a reasonable range.

This is not an argument for artificial balance or equal space for every suggestion. A plainly inapplicable alternative may need only a short explanation. A credible competing approach to a central issue deserves more. As a practical drafting test, consider whether a reader could understand the strongest argument against your conclusion without having to reconstruct it from the other side’s report.

Make the reasoning available to the reader

The Academy’s Model Form of Expert Report distinguishes facts, assumptions and opinions, including opinions taken from other experts. It also emphasises linking conclusions to the supporting text. Those distinctions help readers follow the reasoning rather than simply being asked to accept the author’s assurance that the necessary work has been done.

In practice, it is worth checking the route from each important conclusion back to its supporting material.

• Which observation, calculation or professional judgment carries the conclusion?
• What would change if an assumption proved wrong?
• Does the report explain why apparently contrary material does not alter the answer?

An appendix can contain the underlying detail, but the reader still needs to be shown how that detail affects the opinion.

Care is also needed where the reasoning depends on another discipline. The Model Form distinguishes between opinions of others on which the writer relies in forming their own opinion and those the writer accepts but cannot comment on authoritatively. A clear report should make that dependence visible rather than allowing the reader to assume that every underlying conclusion has been independently assessed by its author.

Carry the analysis into the joint statement

The same care is needed when the experts discuss their differences. CPR 35.12 provides for court-directed discussions and a statement recording agreement and disagreement, with reasons for the latter. The Academy’s guidance on joint statements addresses how those issues and reasons can be presented usefully.

A statement that the experts continue to disagree adds little unless it explains why. The difference may concern an assumed fact, the suitability of a method or the weight attached to particular evidence. Identifying that difference precisely is more helpful than placing two firm conclusions alongside one another. It also gives each expert an opportunity to check whether the disagreement is as extensive as it first appeared.

Before signing a report or joint statement, try reading it from the position of someone who favours the alternative. Have their substantive reasons been addressed, or merely their conclusion rejected? The aim is not to make the expert less decisive, but to make the basis of the decision open to proper examination.

These aspects of presenting and evaluating expert evidence are also covered in the Academy’s The Expert’s Report training.

Read the full judgment in JBX v Frimley Health NHS Foundation Trust. The assessments of the neuropsychiatric and care evidence are at paragraphs 95–97 and 112–117 respectively.

A new Right to Work regime takes effect from 1st October 2026 and may affect experts, their use of associates and, potentially, the terms agreed with instructing solicitors and expert agencies. With civil penalties of up to £60,000 per illegal worker, this is an area where mistakes can prove costly.

The UK government has required employers to carry out checks on employees’ right to work since the Immigration, Asylum and Nationality Act 2006. Section 48 of the Border Security, Asylum and Immigration Act 2025 extends that regime so that, in some circumstances, civil penalty liability can reach beyond the employer who has the direct contractual relationship with the worker.

It is this extension that is particularly relevant to experts, their associates and those who instruct them. The Home Office has issued draft guidance setting out how the new regime is expected to operate.

Associates engaged directly

If you or your expert firm engage an individual associate or subcontractor to help deliver work, you may need to carry out a prescribed Right to Work check before that person starts work.

The position depends on the nature of the arrangement. The new regime is not intended to mean that every self-employed expert or sole practitioner automatically has to be checked. However, where an individual is engaged to help fulfil work which your firm has itself contracted to provide, the arrangement may fall within the extended regime.

The guidance explains the different ways in which a Right to Work check can be carried out. Depending on the circumstances, this may involve a manual document check, a Home Office online check or the use of an approved digital verification service provider. Records may also need to be retained and, where the individual has time-limited permission to work, further checks may subsequently be required.

Associates engaged through another firm

If you or your firm engage associates through their firm or another organisation, there is a separate issue: extended liability.

Under the new regime, liability may in some circumstances extend further up a contractual chain. A business which does not itself employ the individual may nevertheless be exposed if illegal working is later identified.

The guidance provides for a statutory excuse where the relevant requirements are met. These include having appropriate contractual terms in place which require prescribed Right to Work checks to be carried out, control further subcontracting, allow appropriate compliance checks and provide for action to be taken where illegal working is identified. The relevant contractual requirements are set out at page 49 of the current draft Home Office guidance.

The protection is not simply a matter of inserting standard wording into a contract. The arrangements also need to operate properly in practice.

What does this mean for instructing solicitors and expert agencies?

Whilst the immediate focus for experts and expert firms is likely to be the checking of individuals and the review of associate or subcontract terms, the changes may also have consequences for agreements with instructing solicitors and expert agencies.

Those further up the contractual chain may themselves wish to establish a statutory excuse against extended liability and may therefore seek additional contractual assurances from experts about Right to Work compliance, subcontracting and the use of associates.

It would not therefore be surprising to see new Right to Work provisions appearing in expert instructions and agency terms after 1 October.

What should expert firms do now?

Expert witnesses and expert firms that use associates or subcontractors should review their arrangements before the new regime takes effect.

In particular, they should consider:

• whether any individuals they engage may fall within the extended Right to Work regime;
• how Right to Work checks will be carried out where required;
• whether existing associate or subcontract terms need updating;
• whether their contracts permit substitution or further subcontracting; and
• whether clients or agencies are likely to require additional contractual assurances.

The timing is slightly more nuanced than saying that every existing agreement has to be rewritten by 1st October. The new provisions apply to relevant engagements commencing on or after 1st October 2026, and to relevant contractual arrangements entered into on or after that date.

Nevertheless, firms which regularly use associates or subcontractors would be well advised to review their procedures and standard terms now, rather than waiting until a new instruction or engagement arises.

The Home Office guidance is currently in draft and should be checked alongside the legislation before any changes are made to contractual arrangements.

Further information

The Home Office’s current guidance is available in its Draft employer’s guide to right to work checks: 11 September 2026.

This article is intended as a general overview of the changes and does not constitute legal advice.

An expert is perfectly entitled to reach a strong opinion which supports the case of the party instructing them. The difficulty is when the reasoning appears to work the other way round, with the party’s case driving the expert opinion.

That distinction arose in KS v Secretary of State for the Home Department, where First-tier Tribunal Judge Froom considered expert evidence about conditions in Tajikistan. His criticisms ranged from whether the expert had demonstrated the necessary expertise through to the sources relied upon and the extent to which the appellant’s account had been independently examined. By the end of that analysis, the Judge’s view was that the expert was advocating for the appellant rather than providing critical and objective expert evidence.

The limits of an expert’s expertise

The expert described himself as a publicist, PR expert, television presenter and journalist specialising in Russia and post-Soviet states. He held degrees in international law and marketing and had written extensively for newspapers and websites.

The Tribunal was concerned about how much of that experience demonstrated expertise specifically in Tajikistan. Judge Froom noted that the expert did not appear to have visited the country, had held no academic post and listed only one publication, which concerned Russia. He concluded that sufficient expertise on Tajikistan had not been demonstrated.

This is an issue which can easily arise where someone has genuine expertise across a broad subject or geographical area. The question is not simply whether an expert is knowledgeable, but whether their knowledge and experience extend far enough to support the particular opinions they have been asked to give.

Checking the sources

Some of the most striking criticism concerned the source material.

The report stated that honour crimes were deeply ingrained in Tajik culture, particularly in rural areas, and cited four sources. Two dealt with domestic violence against women and the Tajik government’s response, another with religious freedom and the fourth was the Criminal Code of Tajikistan.

“I was not able to find any reference to honour crimes in any of these documents.”

Counsel was invited to identify such a reference and was also unable to do so. Judge Froom described the failure accurately to source the assertions about honour crimes as a “serious flaw in the report”.

The question of sources also arose in SU (Expert reports – format) Bangladesh [2026] UKUT 317 (IAC), which we considered in our article Online sources are not the expert opinion.

In SU, the Upper Tribunal emphasised that material relied upon should be identifiable and capable of being checked, and that experts should go back to the primary source wherever possible. The report must also distinguish between the underlying material and the expert’s own analysis and opinion.

KS illustrates the problem from a slightly different direction. Identifying a source is not enough if the source does not actually support the proposition for which it is being cited. The two decisions therefore fit rather neatly together. In SU, the emphasis was on making the source material transparent and traceable; in KS, the Tribunal was able to trace the sources and found that they did not support an important assertion in the report.

For experts who rely on academic papers, technical literature, industry publications, statistics or other external material, the practical discipline is much the same. The reader should be able to identify what has been relied upon, find it if necessary and understand what the expert says it establishes. A citation does not acquire authority merely by appearing in an expert report.

Assumptions and the client’s account

The Judge was also concerned about the way in which the expert dealt with information provided by the appellant.

Claims had been made about the influence of members of a particular family. The expert did not say that he had investigated those individuals or independently established the extent of their influence. Judge Froom said that the expert:

“simply accepts what the Appellant says and agrees with him”

and concluded that this did not assist the Tribunal.

Expert reports will often depend upon factual assumptions supplied by those instructing the expert. There is nothing unusual about that, and in many cases the expert will have no means of establishing those facts independently. What matters is being clear about the distinction between an assumed fact and something the expert has independently established, and not presenting the former as though it were the latter.

There is an interesting contrast elsewhere in the judgment. Judge Froom identified an issue concerning the alleged police search on which expert evidence might have assisted the Tribunal, but the report did not address it. The report therefore ventured into areas which concerned the Tribunal while leaving unanswered a question on which the expert might have been able to provide useful assistance.

When opinion becomes advocacy

Against that background, Judge Froom said:

“The distinct impression gained is that Mr Kharitonov is simply advocating on behalf of the Appellant”

rather than providing a critical and objective expert opinion. He went on to say that the expert “trespasses into the role reserved to the Tribunal”.

An expert does not preserve their independence by avoiding firm conclusions or carefully balancing every opinion. If the evidence leads strongly in one direction, the expert should say so. Independence lies in the process by which the opinion is reached: considering the relevant material, testing assumptions where appropriate, acknowledging the limits of the expert’s knowledge and allowing the evidence to determine the conclusion.

Seen together, the criticisms in KS explain why the Tribunal became concerned about advocacy. Questions about the expert’s specific expertise would not necessarily lead to that conclusion on their own. Nor would one poorly chosen citation. But where important assertions are unsupported by the sources cited, the client’s account is accepted without sufficient examination and opinions move into matters for the Tribunal, the cumulative effect can be rather different.

Judge Froom attached no weight to the report as expert evidence.

For anyone preparing an expert report, KS is a useful illustration of why independence has to be apparent throughout the report. It can be seen in the treatment of instructions and assumptions, in the selection and checking of sources, in recognition of the limits of the expert’s own expertise and, above all, in whether the reasoning genuinely leads to the opinion rather than being used to support a conclusion already reached.

Read the judgment: KS v Secretary of State for the Home Department on Find Case Law.

A recent Upper Tribunal decision gives some unusually practical guidance for experts who rely on material found online.

In SU (Expert reports – format) Bangladesh [2026] UKUT 317 (IAC), the Tribunal considered expert reports which drew heavily on publicly available internet material. The decision is particularly useful because it deals with something that now affects experts in many fields: how to use online source material properly in an expert report.

The Tribunal’s guidance is fairly straightforward, but important.

Where an expert relies on material found online, the source should be identifiable and capable of being checked. So far as possible, the expert should use the primary source rather than a secondary report about it. If material is quoted, it should be clearly identified as a quotation, with a link to the source and a translation where appropriate.

That is the sourcing side of the exercise.

But there is a separate point which is perhaps more important.

The material found online is not the expert opinion.

It is source material from which the expert may form an opinion.

The expert’s task is to consider that material, explain its relevance and significance, and then form and express an independent opinion based on it.

Both parts matter.

The underlying source material must be sufficiently transparent for the court or tribunal to understand where it came from and, where necessary, check it. But the report must also show the expert’s reasoning. The court needs to be able to understand how the expert moved from the underlying facts and sources to the opinion ultimately expressed.

It is then for the court or tribunal to decide what weight to give that opinion.

Primary sources matter

One point made expressly by the Upper Tribunal is that experts should, where possible, go back to the original source.

That may sound obvious, but it is increasingly easy to rely on a newspaper article, online summary, social media post or another secondary source which itself refers to an underlying document.

For an expert, that can create unnecessary uncertainty.

If a government statement, NGO report, technical paper, regulatory document or other primary material exists, there is obvious value in identifying and considering that material directly.

The Tribunal also stressed the importance of making quotation clear. Material copied from elsewhere should not appear in the report as though it were the expert’s own words.

That is not simply a matter of presentation. It enables the reader to distinguish between:

• the underlying source;
• the facts or information derived from it;
• the expert’s analysis of that information; and
• the expert opinion.

Those distinctions are central to a properly reasoned expert report.

The Academy’s own guidance on the Model Form of Expert Report reflects the same need to distinguish clearly between matters of fact and opinion.

Applying the material to the case

The Tribunal also made clear that a report should do more than set out general material.

Where the expert has been asked to consider the particular circumstances of a party, the report should apply the expert’s general conclusions to those particular facts.

That is an important part of what makes the evidence expert evidence rather than simply research.

A court can often read source material for itself. The value of the expert lies in the specialist analysis and opinion brought to that material.

That may involve explaining technical significance, identifying limitations, placing information in context, drawing on professional experience or explaining how apparently conflicting material should be understood.

The Academy’s What is an Expert Report? guidance similarly stresses that an expert report is not merely a statement of conclusions. The factual basis and reasoning leading to the opinion matter.

Responding to criticism

There is another useful practical point in SU.

Where one of the parties has raised concerns about an aspect of the expert report, the Tribunal said the expert should address those concerns directly, bearing in mind the overriding duty to assist the Tribunal on matters within the expert’s expertise.

That matters because an expert’s duty is not to defend the report at all costs.

If a legitimate criticism identifies an error, omission, weak source or assumption which needs reconsideration, the expert should deal with it.

The Academy’s Code of Practice for Experts places independence, impartiality, objectivity and integrity at the heart of expert practice.

A practical consequence of the online world

The volume of information available to experts has changed enormously.

In many fields, an expert can now locate hundreds of sources in a relatively short period. That can be extremely useful, but volume is not the same as quality.

The discipline lies in showing:

• what material has actually been relied upon;
• where it came from;
• why it is relevant;
• what the expert makes of it; and
• how it contributes to the opinion expressed.

That is why traceability matters.

The court or tribunal does not simply need an expert who can find information. It needs an expert who can analyse that information, explain it and give a properly supported opinion within their expertise.

That is the useful lesson from SU (Expert reports – format) Bangladesh.

Decision:
https://tribunalsdecisions.service.gov.uk/utiac/2026-ukut-00317

BAILII:
https://www.bailii.org/uk/cases/UKUT/IAC/2026/317.html

Related Academy resources

• Model Form of Expert Report
• What is an Expert Report?
• Code of Practice for Experts
• The Expert’s Report training

The High Court’s judgment in Various Claimants v Mercedes-Benz Group AG and others [2026] EWHC 1753 (KB) is a useful reminder that technical expertise, however impressive, is only part of the job of an expert witness.

The litigation involved a huge amount of highly technical expert evidence across a number of disciplines. What is particularly interesting is the contrast in the judge’s assessment of the experts themselves.

Some were praised for the quality and fairness of their evidence. Others attracted criticism over aspects of their compliance with Part 35.

That contrast matters because the experts involved were not lacking in technical expertise. They were experienced specialists dealing with complex issues in major litigation.

The point is rather that being an expert in your field and being an effective expert witness are not quite the same thing.

Under CPR Part 35, an expert’s role is to assist the court on matters within their expertise. That duty overrides any obligation to the person instructing or paying them.

Practice Direction 35 goes further. Expert evidence should be the independent product of the expert, uninfluenced by the pressures of litigation, and experts should consider all material facts, including those which might detract from their opinions.

Those requirements affect far more than the declaration at the end of a report.

They affect how the expert approaches the evidence in the first place.

An expert needs to be able to distinguish between what they know, what they have been told and what they have inferred. They need to consider material which may not support the position of the party instructing them. They need to explain the basis for their conclusions and, where appropriate, acknowledge that there may be a legitimate range of professional opinion.

Perhaps most importantly, they need to remain willing to reconsider their conclusions as the evidence develops.

That may sound elementary to experienced expert witnesses. But the significance of the Mercedes-Benz judgment is that these issues arose in extremely sophisticated litigation involving highly experienced specialists.

The technical complexity of the subject matter did not lessen the importance of the expert’s procedural and professional duties.

If anything, it made them more important.

Where the underlying issues are difficult for a court to evaluate without specialist assistance, the court is particularly dependent upon the expert presenting that specialist knowledge independently, fairly and in a way that can be understood and tested.

That is why the quality of expert evidence cannot be measured simply by the expert’s qualifications or standing in their profession.

The court also needs to be able to trust the process by which the opinion has been reached.

The Academy’s guidance on party-appointed experts reflects this. The expert’s role is to provide unbiased opinion within their expertise and not to assume the role of advocate.

The Academy’s Code of Practice for Experts similarly places independence, impartiality, objectivity and integrity at the centre of expert practice.

And because much of the expert’s evidence will ultimately be contained in the report, the Academy’s Model Form of Expert Report is designed to help experts present their reasoning in a form which assists the court. The Academy’s The Expert’s Report training also deals specifically with compliance, professional guidance and the practical preparation of expert evidence.

The broader lesson from the Mercedes-Benz litigation is fairly simple.

An expert witness brings two kinds of expertise to court: expertise in their professional field, and expertise in performing the role of an expert witness.

The first does not automatically provide the second.

Judgment

A High Court decision provides a useful warning for professionals who become expert witnesses in matters in which they have already been involved.

In MSA Properties Ltd v The Drapery Northampton Ltd [2026] EWHC 1420 (KB), the court had to resolve a dispute concerning deteriorated structural joists in a building in Northampton. Both parties relied on structural engineering evidence.

The point of wider interest for experts concerns the defendant’s expert, who had already been involved in the matter before the litigation.

He inspected the building in 2023 and produced a report for the defendant. That report was not prepared as expert evidence for the proceedings. By 2025, however, he was acting as the defendant’s expert witness.

That change of role became important.

The judge noted that the later expert report did not refer to the earlier inspection or the report produced at the time, despite the expert accepting in oral evidence that the earlier inspection had given him a better view of the joists.

The court was also critical of the extent to which he had engaged with the available evidence and of the opportunities available to him to inspect the property more fully.

The most significant criticism was the judge’s observation that his approach reflected:

“some conflation of his role as a contractor engaged in negotiations with the Claimant and his duties as an expert witness”

There was no suggestion that the expert had deliberately acted partially. The concern was rather that aspects of his earlier role appeared to have carried over into the way he approached his later expert evidence.

That is a useful warning for any professional who changes role during the life of a dispute.

Previous involvement does not automatically prevent someone from later acting as an expert witness. In some cases, that earlier involvement may mean they have valuable knowledge of the subject matter.

But the nature of the role changes.

A contractor, consultant or adviser may previously have been helping a client solve a problem, negotiate a position or achieve a particular commercial outcome. An expert witness has a different duty.

Under CPR 35.3, an expert’s duty is to help the court on matters within their expertise. That duty overrides any obligation to the person from whom they have received instructions or by whom they are paid.

That is not simply a formal statement to be repeated in the declaration at the end of a report. It has practical consequences for the way an expert approaches the evidence.

Transparency about previous involvement

Where an expert has already inspected the subject matter, produced reports, advised a party, taken part in negotiations or expressed views before being instructed as an expert witness, that history may be material to the opinion now being given.

The point is not that every earlier conversation or document must automatically be reproduced. It is that relevant prior involvement should not be left invisible if it bears on the expert’s present reasoning.

That is particularly important because CPR 35.10 requires an expert report to state the substance of all material instructions, whether written or oral, on the basis of which the report was written. Practice Direction 35 also requires the expert to make clear which facts are within their own knowledge and which have been provided by others.

In this case, the earlier involvement was not peripheral. The expert had inspected the joists at a time when they were more fully exposed and had produced an earlier report. He later accepted that the earlier inspection had given him a better view of the joists, yet neither the inspection nor the earlier report appeared in his expert report.

That is a useful illustration of why disclosure matters.

If earlier involvement has informed what the expert knows, assumes or concludes, the court should be able to see that route clearly. If it only emerges later, particularly in cross-examination, the omission can create an avoidable question about the completeness and transparency of the expert’s reasoning.

For an expert changing role, a sensible question at the outset is:

What have I already done, seen, said or concluded in this matter, and what of that is material to the opinion I am now giving?

If someone changes role during the life of a dispute, earlier assumptions may need to be revisited. Previous opinions should not simply be carried forward without being tested against the evidence now available. Earlier inspections, reports and involvement may themselves be relevant and should be dealt with openly where necessary.

The wider framework of the Civil Procedure Rules matters as well.

The overriding objective in CPR 1.1 is to enable the court to deal with cases justly and at proportionate cost. Expert evidence forms part of that process. It is therefore not enough for an expert merely to possess relevant expertise. The evidence has to assist the court in a way which is independent, properly reasoned and proportionate to the issues in dispute.

That is why a change in role can be important.

A professional may quite properly have spent months or years advising a client, negotiating with another party or working towards a particular practical outcome. Those activities may require judgement, persuasion and a close identification with the client’s interests.

The expert witness role is different.

Once instructed as an expert, the professional must be able to step back from that earlier involvement and approach the evidence independently. That may mean reconsidering previous views, acknowledging matters which do not support the client’s position and making clear where earlier involvement has informed the present opinion.

• The Academy’s guidance on party-appointed experts explains the overriding duty owed by an expert to the court.
• The Academy’s Code of Practice for Experts places independence, impartiality, objectivity and integrity at the heart of expert practice.
• The Academy’s Model Form of Expert Report is also relevant where an expert has had previous involvement, because the report should make clear the factual basis of the opinions expressed and the material relied upon.

The lesson from MSA Properties is not that previous involvement disqualifies an expert. It is that a professional who moves from contractor, adviser or negotiator to expert witness must make that transition consciously and transparently. The court may look closely at whether the expert has genuinely separated the new role from the old one, and whether the report gives a clear account of any earlier involvement which bears on the opinion now being expressed.

For experts, that is a useful point to consider before accepting instructions in a matter in which they have already been involved.

See the judgment here on BAILII.

Anthropic has just announced that Claude will start leaving an invisible mark in the text it generates, in an announcement that may have unintended consequences for expert witnesses.

Supported Claude models will embed machine-readable watermarks into generated text. Anthropic says those marks will travel with the words when copied and pasted into other documents and may survive subsequent editing.

The immediate purpose is provenance. As the volume of AI-generated material increases, watermarking offers a potential means of identifying content that has passed through a particular AI system.

For expert witnesses, however, that raises a rather more difficult question: if an expert report contains an AI watermark, what exactly does that prove?

Perhaps less than might first appear.

Detection is not the same as authorship

One of the most important qualifications comes from Anthropic itself.

Anthropic cautions that detecting its watermark does not establish that Claude wrote or originated the material. Rather, it indicates that the content may have been processed by Claude.

That distinction could be particularly important in expert evidence.

An expert might write an entire report themselves and subsequently ask an AI tool to correct spelling and grammar. Another might use AI to improve the structure of material they have already written. A third might provide their analysis and ask an AI system to turn it into polished prose.

Those activities are quite different from asking an AI system to examine the underlying evidence, perform the analysis and suggest the opinion the expert should express.

Yet all involve AI-generated or AI-processed text.

Consider four possible uses:

• An expert writes their report and uses AI to check spelling and grammar.
• The expert writes the substance but asks AI to improve its structure or expression.
• The expert provides their analysis and asks AI to draft passages expressing it.
• The expert asks AI to analyse the evidence and assist in formulating the opinion.

All involve AI, but they plainly do not raise the same questions about the independence of the expert’s analysis and opinion.

That makes the presence of a watermark potentially useful evidence, but not necessarily evidence of what might initially be assumed.

Provenance of the text is not provenance of the opinion

This is perhaps the central issue for expert witnesses. A watermark may provide evidence about the provenance of the words. It might indicate that Claude, or another AI system using similar technology, has been involved in generating or processing them:

• It cannot, by itself, establish the provenance of the opinion.
• It does not tell us who examined the evidence.
• It does not tell us who identified the relevant issues.
• It does not tell us whether the expert carried out the underlying calculations or analysis.
• And, most importantly, it does not tell us whether the conclusion was independently reached by the expert or suggested by an AI system.

The same paragraph could conceivably carry an AI watermark whether Claude corrected three grammatical errors in it or generated the paragraph from scratch.

For expert evidence, those are very different propositions.

The absence of a watermark proves little too

There is also a problem in the opposite direction.

No watermark does not necessarily mean no AI.

Anthropic acknowledges that detection can become more difficult where generated material has subsequently been heavily edited, paraphrased, translated or combined with other text. A passage may also simply be too short to provide sufficient information for reliable detection.

This creates an important limitation if watermark detection ever begins to be used when examining expert reports.

A positive result may indicate AI involvement without establishing the extent or nature of that involvement whilst a negative result cannot necessarily establish that there was no AI involvement at all.

Watermark detection should therefore not be treated as a binary test of whether an expert “used AI”.

A new question in cross-examination?

The development also raises an interesting practical possibility.

“Your report contains a watermark indicating that it was processed by Claude. How did you use it?”

The answer might be entirely unremarkable.

The expert might explain that they wrote the report themselves and used Claude to check its grammar. They might explain that they used it to condense a lengthy passage they had already written. Alternatively, the answer might reveal that AI played a much greater role in analysing material or formulating the opinions contained in the report.

The takeaway is that whilst the watermark itself cannot distinguish between those situations, it may, however, provide the starting point for questions which can.

That means the increasingly important question may not simply be:

“Did you use AI?”

It may instead be:

“Precisely how did you use it?”

Watermarking is becoming a real technology

Anthropic is not alone in pursuing this approach.

Google has already developed SynthID Text, a watermarking technology for AI-generated text. Rather than inserting a visible label or conventional metadata, SynthID subtly affects the model’s choice of tokens as text is generated. Across a sufficiently long passage, those choices create a statistical signature which can subsequently be detected.

Anthropic has not publicly confirmed that Claude’s watermark uses the same technical method, so the two systems should not be assumed to work identically.

The significance is broader. Text watermarking is moving beyond academic research and into mainstream generative AI products.

If the technology becomes widespread, lawyers, courts and experts may increasingly encounter claims that a document has been identified as having passed through an AI system.

Understanding what such a finding does, and does not, establish will therefore become important.

The expert remains responsible

The Academy’s guidance on the use of Artificial Intelligence by expert witnesses already addresses the more fundamental issue.

AI can be a useful tool, but it cannot substitute for the expert’s own expertise, analysis and opinion. Experts remain responsible for the evidence they give. Where AI has played a significant role, keeping appropriate records of how it was used may also become increasingly important, particularly if the expert is subsequently required to explain that use in court.

Watermarking adds another dimension to that responsibility.

It may make some uses of AI more readily identifiable. But identifying that an AI system touched the words is not the same as establishing who produced the intellectual work behind them.

For expert evidence, that distinction is crucial.

Provenance of the text is not necessarily provenance of the opinion.

As AI provenance technology develops, the ability of an expert to explain what the AI did, what the expert did, and where the opinion actually came from may become increasingly important.