In partnership with
This week a judgment from the Upper Tribunal spelled out what many practitioners have been quietly assuming: uploading confidential client documents to a public AI tool like ChatGPT breaches client confidentiality and waives legal privilege. Simmons & Simmons responded with a practical privilege guide, Harvey announced it is embedding in Microsoft 365 Copilot, and the CJC consultation on AI in court documents picked up further commentary.
AI in Practice
Tribunal says uploading to ChatGPT waives privilege
In Munir, R (On the Application Of) v Secretary of State for the Home Department (AI hallucinations; supervision; Hamid) [2026] UKUT 81 (IAC), the Upper Tribunal (Immigration and Asylum Chamber) made what is believed to be the first explicit judicial statement in England and Wales that uploading confidential documents to a public AI tool waives legal privilege.
At paragraph 60, the tribunal stated:
“Uploading confidential documents into an open-source AI tool, such as ChatGPT, is to place this information on the internet in the public domain, and thus to breach client confidentiality and waive legal privilege, and any such conduct might itself warrant referral to the regulatory body and should, in any event, be referred to the Information Commissioner’s Office.”
It added that any such conduct might warrant referral to a regulatory body and should, in any event, be referred to the Information Commissioner's Office.
The case itself joined two matters before the tribunal. One involved a solicitor who had included a non-existent case in grounds of appeal; most likely the result of an AI hallucination from an unverified search result. The broader context is supervision, as we have said before. The tribunal was concerned not just with the use of AI but with the absence of adequate checks on AI generated output.
It moves the privilege point from professional guidance into case law. Until now, the risk of privilege waiver from using public AI tools has been discussed in SRA guidance, Bar Council ethics notes and firm level policies, but it has not been the subject of a direct judicial statement in a reported decision. The tribunal's language is unambiguous: public AI tools and privileged material do not mix.
It is worth noting that this is a first-instance tribunal decision, and therefore non binding. But the reasoning is clear and the language is strong so this author expects it will be cited and is certainly persuasive. Firms that have not yet addressed this in their AI policies should do so now.
Read: BAILII
On your radar
Simmons & Simmons launches AI and legal privilege guide: The firm has published a free guide and policy framework designed to help organisations protect legal privilege when staff use AI systems. It covers the distinction between open and closed AI systems, key privilege risks, and includes a pro forma staff policy that organisations of any size can adapt. The timing, shortly after the aforementioned Munir judgment, is not coincidental. Why it matters for UK lawyers: if you need a starting point for an internal AI privilege policy, or want to benchmark your existing one, this is a practical, ready-made resource from a firm with a serious AI practice. (Artificial Lawyer / Simmons & Simmons)
Harvey to integrate with Microsoft 365 Copilot: Harvey announced this week that it is building an integration with Microsoft 365 Copilot, launching in Q2 2026. Lawyers will be able to @Harvey in Copilot to ask legal questions, research issues, or analyse documents without leaving their existing Microsoft workflow. When a question becomes a larger task, users can click through to Harvey for deeper analysis. Why it matters for UK lawyers: for firms already on Harvey and Microsoft 365, this removes the context-switching between platforms. It also shows that the legal AI market is consolidating around embedded workflows rather than standalone tools. (Legal IT Insider)
Pinsent Masons comments on the CJC AI consultation: Following last week's lead story on the Civil Justice Council's proposals for mandatory AI declarations in witness statements, Pinsent Masons has published its own commentary. The consultation closes on 14 April 2026. Why it matters for UK lawyers: if your firm is preparing a response, or advising clients on theirs, the law firm commentary now appearing is useful for testing your own position. The CJC's proposals remain the most significant rule-making signal from the judiciary on AI use in English civil proceedings. (Pinsent Masons)
August launches real-time contradiction detector for depositions: US legal AI platform August has launched Live Assist, a tool that joins live conversations, cross-references what is being said against uploaded case documents, and flags contradictions in real time with source citations. Why it matters for UK lawyers: this is a US product and the immediate use case is depositions, which are not a feature of English litigation. But the underlying capability (real-time contradiction detection against a document set) has obvious applications for witness preparation, client interviews and settlement negotiations under any legal system. Worth watching as the technology matures. (Artificial Lawyer)
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 and actual products that I have tried, with some brief thoughts from me.
I remain a firm fan of Wispr Flow and am flattered they continue to support the newsletter. I am grateful to readers of recent weeks for taking a look at the tool and I hope you are finding it suits your productivity. Once again, this newsletter was dictated in part with this tool!
If you have not tried Wispr Flow, I recommend their free trial.
Dictate prompts and tag files automatically
Stop typing reproductions and start vibing code. Wispr Flow captures your spoken debugging flow and turns it into structured bug reports, acceptance tests, and PR descriptions. Say a file name or variable out loud and Flow preserves it exactly, tags the correct file, and keeps inline code readable. Use voice to create Cursor and Warp prompts, call out a variable like user_id, and get copy you can paste straight into an issue or PR. The result is faster triage and fewer context gaps between engineers and QA. Learn how developers use voice-first workflows in our Vibe Coding article at wisprflow.ai. Try Wispr Flow for engineers.
For Review
AI and Legal Advice Privilege Guide (Simmons & Simmons)
A practical policy template and guidance document covering the privilege risks of AI use, with a particular focus on the distinction between open and closed systems. Includes a pro forma staff policy. Freely available and suitable for organisations of any size looking to create or update their internal AI governance.
Read: Simmons & Simmons
"The Mobile AI Gap: The Problem in Legal Tech No One Is Talking About" (Artificial Lawyer)
A survey-backed piece arguing that while 80% of lawyers now use AI weekly, the tools remain overwhelmingly desktop-bound. Only 20% primarily access AI on a smartphone. For lawyers who work across meetings, court and travel, this is a practical gap that limits the value of their existing tools and the article shows that lawyers are using mobile tools more and more. Useful context for any firm evaluating its AI tooling strategy.
Read: Artificial Lawyer
Munir, R (On the Application Of) v SSHD [2026] UKUT 81 (IAC) — full judgment
The full text of the judgment. Worth reading in full if you advise on professional conduct, AI governance, or immigration law. The tribunal's analysis of supervision failures and AI hallucinations is relevant well beyond the immigration context.
Read: BAILII
Practice Prompt
This week's prompt helps you check whether your current AI tools put privilege at risk. Ensure you fill in context and other aspects marked with {}. Remember to adhere to the Golden Rules and do not upload confidential or privileged information to public tools.
I am a {solicitor / barrister / in-house counsel} working in {practice area} at {type of firm or organisation}.
I currently use the following AI tools in my work:
{List each tool, e.g.:
- ChatGPT (free version, via browser)
- Microsoft 365 Copilot (enterprise, via firm licence)
- Claude (Pro subscription, personal account)
- Harvey (firm-wide deployment)}
For each tool listed, produce a table with the following columns:
1. Tool name
2. Open or closed system (i.e., does the provider use inputs for training or make them accessible outside the organisation?)
3. Privilege risk level (high / medium / low) — based on whether uploading client material to this tool could constitute placing it in the public domain
4. Whether uploading privileged or confidential client documents to this tool is safe, unsafe, or requires further investigation
5. Recommended safeguard (e.g., "do not upload client documents", "safe for use under firm data processing agreement", "check enterprise settings with IT")
Then list any immediate actions I should take based on this assessment.
Base your analysis on the principle from Munir, R (On the Application Of) v SSHD [2026] UKUT 81 (IAC) that uploading confidential documents to a public AI tool breaches client confidentiality and waives legal privilege. Flag where you are uncertain and recommend I seek further guidance.
How did we do?
Hit reply and tell me what you would like covered in future issues or any feedback. We read every email!
Finally, a small poll to gauge interest in a future series of articles. Please vote!
What you read a series about implementing AI in a smaller firm, particularly litigation practice?
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!
