This article was originally published in Law360 on 29 June 2026 and is reproduced by kind permission of the publishers.

Both the Bar Council and the Law Society have responded to an interim report and consultation on the use of artificial intelligence for preparing court documents in April and May, respectively.[1]

The consultation was published on 17 February by the Civil Justice Council, an advisory body to the government and judiciary.[2]

Focus on the use of AI by expert witnesses who have been instructed to provide independent expert evidence in English court proceedings is increasing. Whether the current rules are clear or additional rules are needed is a subject open to ongoing debate.

The consultation includes consideration of whether existing experts' duties and guidance are sufficient to cover experts' use of AI. It proposes that experts are subject to a new requirement to explain their use of AI, other than for transcription or other administrative uses, and identify the AI tools used.

Specifically, the briefing considers:

  • Whether the existing rules concerning expert evidence in English court proceedings are sufficient to mitigate AI-related risks;
  • If further rules are needed;
  • How any further rules might be implemented; and
  • The practical steps that lawyers and experts can take to ensure that AI tools are used appropriately and in accordance with experts' duties to the court.

Experts and generative AI

Generative AI has real potential for expert witnesses in litigation. The possible use cases for generative AI tools for experts are significant — e.g., pattern analysis of large datasets, summarising complicated technical information, and stress testing potential weaknesses in opinions. The consultation expressly acknowledges that an expert's evidence may properly include AI-generated material.

There are also material risks. Risks associated with the use of generative AI are pertinent in particular to the preparation of expert reports, including deepfakes, hallucinations, and inaccurate or biased outputs.

It goes without saying that if expert evidence is inaccurate or biased, then its weight is significantly diminished. In England and Wales, experts also owe important duties to a court to provide opinions that are independent and unbiased, and that only relate to matters within their expertise.

There are also more fundamental issues to consider. The role of experts in litigation is based on the idea that they draw on experience and expertise to provide specialist opinions. Generative AI tools such as large language models, or LLMs, do not think like a human does, or draw on lived experience. Instead, LLMs generate outputs through applying predictive analysis to tokens based on the datasets on which they are trained.

The exact way in which LLMs generate a specific output is not necessarily fully understood, i.e., the so-called black box problem. If an expert uses generative AI, how does this sit with an expert's duties to provide opinions on matters within their expertise?

Another issue is the level playing field: Individual LLMs have different strengths and weaknesses and are not equal to each other, including in terms of architecture and parameters. How does an expert justify which choice of LLM they have made? What happens if some experts have access to the most powerful LLMs but not others?

There are no straightforward answers to these difficult questions, and inevitably it will take time and real-life examples for experts, lawyers and the judiciary to grapple fully with them. However, in the absence of developed rules and settled policy at present, it is instructive to return to first principles.

Accordingly, an expert report must reflect the expert's independent view rather than being overly reliant on the output of an AI tool. On a practical level, an expert must also be able to defend their opinion when cross-examined.

Existing duties and use of AI

The regulation of expert evidence in English court proceedings is contained in Civil Procedure Rule 35 and Practice Direction 35, and the guidance issued by the Civil Justice Council.

These were introduced before the explosion in AI technology, but the scope of certain duties would require an expert to explain the extent they have relied on AI in preparing their report and ensure their report does not contain hallucinations or inaccurate material and the opinions expressed are their own.

For example, Practice Direction 35 paragraph 3.2(2) requires an expert to "give details of any literature or other materials which have been relied on in making the report."

Meanwhile, paragraph 3.2(5) requires an expert to "say who carried out any examination, measurement, test or experiment which the expert has used for the report ... and [whether they have] been carried out under the expert's supervision."

Further, an expert report must include a statement of truth which incorporates a declaration that the opinions expressed represent the expert's true and complete professional opinions on the matters to which they refer.

If an expert fails to explain their reliance on AI to generate substantive content for their report or to undertake analysis on which they base their opinion or fails to verify any AI output or ensure the report reflects their independent opinion, in principle they will be in breach of their duties under the rules.

However, given the growing number of instances when inaccurate AI-generated material has been included in various documents submitted to court, consideration is being given to expanding these requirements.

At present, the debate is focused on whether experts should be subject to an express requirement to disclose their use of AI. Beyond transparency requirements, new guidelines on what is and is not appropriate use of AI technology are likely to be considered to assist experts in ensuring that they meet their duty of providing an accurate and independent opinion.

Expert evidence must be useful and reliable for the courts, otherwise it does not serve its purpose. Experts are drawn from a wide range of backgrounds with varying levels of experience of litigation, and clarity over AI use is key.

Current proposals for rulemaking

A potential approach raised in the consultation is that experts disclose how AI has been used. In the consultation, the Civil Justice Council proposed that the statement of truth in an expert report includes a new requirement to explain what use of AI has been made — other than transcription and/or administrative uses — and identify the AI tools used.

The consultation draws a distinction between the use of AI for administrative purposes such as transcription and spell-checking, and its use to generate evidence.

A similar distinction is seen in the Academy of Experts Guidance for Expert Witnesses on the Use of Artificial Intelligence, published in January,[3] which compares high-risk uses, such as to generate substantive content or undertake material analysis on which their opinion will be based, and low-risk ones, such as to summarise or extract data on which the expert will not rely.

However, it can be difficult to identify where to draw the line. Not all use of AI by experts carries the same level of risk. Disclosure requirements should not be overly burdensome in capturing uses where there would no practical benefit to require disclosure.

There is also a risk that the concept of disclosure of use of AI becomes less meaningful as AI tools become integrated into everyday enterprise business software.

In its response to the consultation, the Law Society agreed that disclosing the use of AI should not be necessary for straightforward activities such as spell-checking. However, it noted there is a growing body of evidence showing use of AI for even transcription and translation carries with it a wide range of risk — therefore arguing such use should be disclosed.

What would a declaration achieve in any case? In its response to the consultation, the Bar Council noted that a declaration that an expert has used AI does not mean that they have used AI in a manner which is consonant with their duty as an expert. The usefulness of disclosure requirements also turns on the level of detail disclosed — would a line in a report noting that AI had been used suffice?

Arguably a better approach is to introduce new guidelines that clarify how the use of AI affects experts' duties, including addressing the complexity around defining the types of AI use.

The Bar Council suggests that new guidance could be tailored to disciplines in which experts are commonly consulted and could provide examples of common AI use cases that ought to be declared, which could be added to over time.

Guidelines could be included in an expanded Practice Direction 35 or contained in new guidance from the courts which could also explain and give examples of appropriate and inappropriate use of AI by experts with the aim of ensuring that reports are accurate, independent and satisfy the experts' duties.

The Law Society also recommended the introduction of court guidance providing red lines on when an expert can use AI with case examples, in addition to amendments to the statement of truth.[4]

But what about the more fundamental issues concerning experts and AI such as the level playing field? It is fair to say that at present, the debate has yet to develop to the stage of potential rulemaking.

As technology tends to move faster than rulemaking, inevitably we are likely to see in coming months and years examples of what experts should and should not do as regards AI reported in real cases, which will serve to test the existing rules, any new rules and drive further policy making.

Practical steps

Lawyers can take steps now when working with experts to understand the types of risk of any proposed use of AI and actions that can mitigate those risks and assist them in relation to their duties.

Lawyers should address the use of AI by the expert in the engagement letter. In some cases, an expert may be restricted from using AI, for example on the client's instructions. This should be made clear from the outset.

Where AI is permitted, you should flag the key considerations, such as that the expert must explain its use, independently stand behind the conclusions of their report and cannot rely on AI generated answers.

The expert should ensure that they identify their proposed AI use to the lawyers in advance, so that sufficient safeguards can be put in place. Consider the expert's knowledge and experience and ensure they understand the risks in the context of litigation.

Published in February, the U.K. Upper Tribunal decision in U.K. v. Secretary of State, for example, provided guidance that inputting information to open-source AI tools is akin to disclosing it to the world and will permanently waive confidentiality and privilege over such documents.[5]

Relevant guidance, such as that of the Academy of Experts, should be identified. While this guidance does not have official status in the English courts, it sets out practical steps and a checklist that experts could follow when they plan to use AI in their work.

These include implementing processes to ensure adequate human oversight and documenting key decisions about the use of AI and keeping a record of key steps such as how the tool was used, how the output was used and the safeguards implemented.

Conclusion

Generative AI has significant potential to support experts' analysis, but also presents material risks that if not controlled, could undermine the weight of expert evidence and potentially cause experts to breach their duties to the court. The use of generative AI also raises fundamental issues about the nature of expert evidence and level playing fields.

It is inevitable that rules and guidance around expert evidence will need to evolve, and that new rules and guidance will need to withstand being tested in the real world. The current debate is focused on clarifying how experts can use AI in accordance with their duties and transparency requirements. As the courts grapple with real world issues, new areas of debate will emerge.

[1] https://www.barcouncil.org.uk/static/bbb74182-4859-4a76-85b1e4b729b84489/Bar-Council-response-to-the-CJC-consultation-on-AI.pdf; https://www.lawsociety.org.uk/campaigns/consultation-responses/use-of-ai-for-preparing-court-documents

[2] https://www.judiciary.uk/wp-content/uploads/2026/03/Interim-Report-and-Consultation-Use-of-AI-for-Preparing-Court-Docume.pdf

[3] https://academyofexperts.org/practising-as-expert/expert-witness-guidance/guidance-for-expert-witnesses-on-the-use-of-artificial-intelligence-ai/

[4] At para 130.

[5] U.K. v. Secretary of State for the Home Department [2026] UKUT 81: https://www.nortonrosefulbright.com/en/inside-disputes/blog/202604-court-guidance-that-use-of-open-source-ai-waives-confidentiality-and-legal-professional

Recent articles

  • content-type:insights/blog
  • content-type:generic
  • nrf-migrated:article-categories/inside-disputes/type/blog-post