Key takeaways
Witness statements: test is now whether AI ‘generated or influenced’ the content
Proposed declaration asks whether AI shaped substance of evidence, not just whether it wrote the words.
Experts must explain AI use, and may have to name the tool
A new declaration is recommended wherever AI generates or influences the substance of an expert's report.
No AI declarations for lawyers, but accountability will be written into the rules
A new Practice Direction will set out who is responsible and the sanctions that apply for misuse.
In April, we looked at the Civil Justice Council's (CJC) Interim Report and Consultation on the Use of AI for Preparing Court Documents. The CJC has now published its final report, dated 29 October 2026.
The report makes recommendations only. They are addressed to the Civil Procedure Rule Committee (CPRC), the CJC and the legal services regulators, and don't change any rules by themselves. The CJC will review progress on its 15 recommendations in its next Annual Review. The direction is now clear.
The approach remains targeted and risk-based. The CJC's starting point is that regulating AI use in court documents ‘should remain with legal representatives and their professional regulators’, supported by a Practice Direction.
Witness statements
The proposed declaration has changed in an important way. The interim report proposed a declaration that AI had not been used to generate, edit, rephrase or embellish trial witness evidence. The final report recommends that a witness of fact declares that ‘AI used in its preparation has not generated or influenced its substantive content.’
We understand ‘Generating’ to mean producing or modifying content. We understand ‘Influencing’ to mean using an AI tool that affects how content is selected, analysed, developed or evaluated, even without producing it.
This goes further than asking whether AI wrote the words. A tool that helps decide which events to include, or how to structure a witness's account, could ‘influence’ it even if every word is the witness's own. Firms should review their protocols, including any use of summarisation or chronology tools. The declaration doesn't ban any tool; it ‘focuses on the result of such use, not on the use itself or the tool that produces the result.’
Two points are new. The declaration would apply whether or not the party has a lawyer. And no new rule is recommended for statements prepared and signed by a legal representative who is not a witness of fact, the lawyer remains responsible for their accuracy.
Expert reports
The CJC's position on experts is broadly unchanged but firmed up. It recommends a new declaration wherever AI generates or influences the substantive content of an expert's report. This must give enough information for the methodology and the AI use to be scrutinised, including, where necessary, disclosing the tool to the other side and the court.
Because expert evidence must be open to testing, the CJC prefers a separate obligation to the existing methodology disclosure under CPR PD 35. Good-practice guidance will amend the revised 2014 Guidance for the Instruction of Experts in Civil Claims.
Statements of case and skeleton arguments
As in the interim report, no AI declaration is recommended for statements of case, skeleton arguments or other advocacy. However, the Practice Direction should make clear that the legal professional responsible for the document must be identified.
It would also consolidate the case law on AI use and on sanctions for misuse. Our message from April still applies: hallucinated citations and other AI-generated errors remain the legal representative's responsibility. The difference is that expectations and consequences will be set out in one place.
Disclosure, translation and permission
Disclosure: Subject to the Disclosure Review Working Group's work, no reforms are needed. The privilege questions we raised in April remain worth considering.
Translation: All translations should be certified by a professional human translator, and AI translations should come with the original text.
Permission: Parties won't need the court's permission to use AI, or to give details of the tools used.
Litigants in person and professional guidance
As we anticipated in April, the CJC will now urgently consider AI use by litigants in person. The regulators are also asked to explore joint guidance and to build responsible AI use into training.
What this means for you
AI has the potential to bring real efficiency and support litigation in a range of ways, provided it is used carefully and with the right safeguards. The final report makes clear those safeguards matter most in evidence, not advocacy, and any use of AI must be subject to appropriate human oversight, review and professional judgment. In-house teams can start by:
checking whether any AI step in preparing witness evidence affects how content is selected, analysed or worded
agreeing with experts, at instruction, how AI will be used and recorded
ensuring a named, responsible lawyer for every AI-assisted document, reviews any AI-generated output and keeps a record of prompts used and steps taken to ensure the accuracy of the final document.
Find out more about our Commercial Dispute Resolution team or contact us today to discuss how we can support you.

