In partnership with

The question the profession has been asking for a year now has an interim answer, and it is not the one many were hoping for. This week the Civil Justice Council set out where its thinking has landed on AI in the preparation of court documents, and the short version is that your existing professional duties are being treated as sufficient. In the same week the Crown Prosecution Service apologised for putting two non-existent cases in front of the High Court, which is a fairly pointed reminder of why the duty to check matters more than any new rule.

AI in Practice

The Civil Justice Council's interim answer: your existing duties already cover AI

The Civil Justice Council has published an update on the responses to its consultation on the use of AI in preparing court documents. The consultation window has now closed, and the Working Group expects to publish its final report later in 2026, so this is a signal of the direction of travel rather than a settled position. It is worth reading, because it answers the question this newsletter flagged as one to watch after last week's piece on AI "slop": what, if anything, the courts intend to do about AI-generated material.

On the central question, the responses show a high degree of alignment. For legal professionals preparing pleadings, skeleton arguments, advocacy documents and the like, the consensus is that no additional formal requirements relating specifically to AI are currently needed, because the existing professional responsibility frameworks are considered sufficient to govern how these tools are used. In other words, the duty to check that a case exists and says what you claim it says is not a new AI duty, it is the same old duty applied to a new tool.

The one area still genuinely open is witness statements. Respondents disagreed on whether extra safeguards or disclosure requirements are needed, and the recurring concern was preserving the authenticity, integrity and personal recollection of the witness, which is precisely what a fluent AI draft can quietly erode. The Council is exploring whether some proportionate approach, possibly including a form of disclosure, would help rather than complicate matters. Expert evidence is heading towards proportionate transparency, and litigants in person are being treated as a distinct and evolving challenge, which connects directly to last week's lead and to the two items in this week's radar.

This writer's reading is that the profession asked for a rulebook and is being told, politely, that it already has one. This writer thinks that is the right call for professional drafting, but it leaves the real work exactly where it has always sat: in supervision, in verification, and in the firm's own policy on how AI is used and recorded. A firm hoping for a bright-line rule to point fee earners at will be disappointed. The practice prompt at the end of this issue is built to help you check your own drafting policy against where the Council appears to be heading.

On your radar

  • The CPS apologises for putting fake AI cases before the High Court: The Crown Prosecution Service has admitted that two of the authorities it relied on in an extradition appeal did not exist, after Mr Justice Sweeting found the citations were non-existent. The false cases appeared first in the CPS's grounds of opposition and were likely to have originated from AI, and the critical failing, on the CPS's own account, was that the reviewing lawyer did not check the document before it was filed. The CPS apologised, reviewed almost 80 other cases handled by the same lawyer, and reinforced its verification processes, and the judge accepted that the error had no effect on the outcome. Why it matters for UK lawyers: when a national prosecuting authority files hallucinated citations, no firm can treat this as someone else's problem, and the failure point was a missing human check rather than the tool. Confirm that every court document your team files passes through a citation-verification step that is written down and actually followed. (Law Society Gazette)

  • The same AI-drafting pattern is now affecting regulators, not just courts: The Independent Office for Police Conduct reports that at least a quarter of the judicial review requests it receives are likely to have had AI assistance, with some quoting legislation that does not exist in England and Wales. Its review requests rose 24.3% to 3,051 in 2025/26, while the proportion of complaints upheld fell from 39.5% in 2022/23 to 26.8%, which suggests fluent AI drafting is generating more volume without improving the underlying merits. Why it matters for UK lawyers: the burden of sorting real arguments from invented ones is spreading beyond the courts to tribunals, ombudsmen and regulators, so any lawyer who handles regulatory complaints or public law work should expect more polished submissions built on non-existent law. Treat a confident citation from an unrepresented party as something to verify, not to trust. (Bracknell News / PA)

  • A reminder that adoption looks different if you will never own the stack: A Law Society Gazette roundtable on AI and mid-sized firms, reported by Joanna Goodman, makes a point that the Kirkland and Shoosmiths headlines can obscure: for most firms the technology cuts both ways, exposing flawed processes rather than fixing them, and the value comes from understanding your own workflows before automating them. Why it matters for UK lawyers: the recent run of "build your own model" coverage does not describe the reality for the overwhelming majority of practices, and the useful question is not which stack to own but which of your processes are sound enough to automate. Pick one repeatable workflow, map how it actually runs today, and fix the process before you point a tool at it. (Law Society Gazette)

Ad Break

In order to help cover the running costs of this newsletter, please check out the advert below. In line with my promises from the start, adverts will always be declared.

Speak naturally. Send without fixing.

Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Not rough transcription you have to clean up. Actual polished text — ready for email, Slack, or any app.

Speak the way you think. Go on tangents. Change your mind mid-sentence. Flow strips the filler, fixes the grammar, and gives you text that reads like you spent five minutes writing it.

89% of messages sent with zero edits. Millions of professionals use Flow daily, including teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.

For Review

Privilege and AI hallucinations (Slaughter and May, The Lens)

A short but genuinely novel piece on a risk almost nobody is discussing: when a firm investigates its own AI-generated error, what protects that investigation? The established cases tell you to check AI output and warn that uploading confidential material to an open tool can waive privilege, but the untested question is whether an internal review into a hallucinated citation is itself disclosable. Worth reading before you have to run one of these reviews rather than after.

AI and litigation: an update on the CJC consultation findings (Civil Litigation Brief)

Gordon Exall's breakdown of the same Civil Justice Council update that leads this issue, written for litigators who want the granular detail rather than the summary. If you have any involvement in your firm's drafting policy or supervision arrangements, this is the version to read closely, because it sets out the reasoning behind each strand of the Council's thinking.

Read or listen: Civil Litigation Brief

Practice Prompt

Try the below prompt to check your firm's policy on AI in court documents against the Civil Justice Council's interim position, which is the practical gap this week's lead leaves open. Ensure you fill in context and constraints and other aspects marked with {}. Remember to adhere to the Golden Rules and do not upload confidential or privileged information to public tools.

You are assisting a knowledge or risk lawyer at a law firm in England and Wales. Your task
is to gap-check the firm's policy on the use of AI in preparing court documents against the
Civil Justice Council's interim direction of travel, as a planning aid only.

Context to apply:
- The current policy: {paste the relevant section of your AI or drafting policy, or describe
  it if there is no written policy}
- Who drafts what: {e.g., "fee earners draft pleadings and skeletons; paralegals draft
  witness statements from client notes"}
- Practice areas and courts: {e.g., "civil litigation, County Court and High Court"}
- Tools in use: {e.g., "a firm-wide assistant" / "public tools, informally"}

Produce a gap analysis under these headings:

1. Pleadings, skeletons and advocacy documents
   The Council's interim view is that existing professional duties are sufficient and no new
   AI-specific rule is needed. Check whether the policy makes verification and supervision
   explicit for these documents, and flag where it relies on assumption rather than a written
   step.

2. Witness statements
   This is the live area, where the concern is preserving the authenticity, integrity and
   personal recollection of the witness. Check whether the policy addresses how AI may and
   may not be used in drafting statements, and whether the firm records any AI involvement.

3. Expert evidence
   The direction is towards proportionate transparency. Check whether the policy says anything
   about AI use in expert reports and how that is disclosed.

4. Litigants in person and inbound material
   Check whether the policy covers how the firm responds to suspected AI-generated material
   received from the other side, including a duty to verify before relying on or answering it.

5. Verification and sign-off
   Identify the single point at which a human confirms every cited authority exists and is
   correctly stated, and flag if that point is missing or informal.

6. Verdict and next steps
   Rate the policy's readiness as LOW, MEDIUM or HIGH against the Council's interim position,
   list the two or three changes that would most improve it, and note what should be revisited
   once the final report is published.

Constraints:
- {Add firm-specific constraints, for example your existing supervision framework or the
  courts you appear in.}
- Apply the law and procedural framework of England and Wales only.
- Do not state the Council's position as settled: it is interim, with a final report expected
  later in 2026. Do not invent rules, cases, or the contents of the final report.
- This is a planning aid, not legal advice, and the supervising solicitor remains responsible
  for the firm's policy and for every document filed.

How did we do?

Hit reply and tell me what you would like covered in future issues or any feedback. We read every email!

Thanks for reading,

Serhan, UK Legal AI Brief

Disclaimer

Guidance and news only. Not legal advice. Always use AI tools safely.

Recommended Newsletters

Below are a few newsletters that I recommend, for various reasons. Check them out!

Staying Ahead with AI

Staying Ahead with AI

Step by step on how to use the latest in AI and how it ranks against what you're already using!

There's An AI For That

There's An AI For That

The #1 AI newsletter. Read and trusted by over 2.5 million readers, including employees at Google, Microsoft, Meta, Salesforce, Intel, Samsung, Zoom, Wix, HubSpot, Nebius, Suno, Zapier, as well as ...

Superhuman AI

Superhuman AI

Keep up with the latest AI news, trends, and tools in just 3 minutes a day. Join 1,500,000+ professionals.

Keep Reading