Suggest a correction

Content

Civil Litigation Update (Jun 2026) - Artificial intelligence and the officer of court

Match the official source exactly.

What needs fixing?

Wrong or mistyped textWrong structure or numberingWrong title or detailsDuplicated content

Proposed text

Academy Publishing is a division of the Singapore
Academy of Law (“SAL”).
 SAL is the promotion and development agency for
Singapore’s legal industry.
Its vision is to make
Singapore the legal hub of
Asia. It aims to drive legal excellence through developing thought leadership, world-class infrastructure and legal solutions. It does this by building up the intellectual capital of the legal profession by enhancing legal knowledge, raising the international profile of
Singapore law, promoting
Singapore as a centre for dispute resolution and improving the efficiency of legal practice through the use of technology. More information can be found at www.sal.org.sg.
CIVIL
LITIGATION
UPDATE
(2026) CLU 1
 — Jun 2026 Issue 1
Civil Procedure
Evidence
June 2026
by Jeffrey PINSLER
ARTIFICIAL INTELLIGENCE AND THE OFFICER OF
COURT
A
Introduction
B
Objective test of reasonableness applies to presentation of authorities
C
Using AI-generated information responsibly
D
What is at stake
E
Responsibility of law practices
F
Sanctions
G
Ministry of Law's Guide for using generative AI in the legal sector

2
A
Introduction 1
There was a time when most lawyers regarded a breach of the duty to the court to present true and accurate information as a fundamental ethical violation and a betrayal of the lawyer’s responsibility as an officer of the court. Until recently, it was rare for a lawyer to be held accountable for misleading the court on the law. Yet, between September 2025 and early 2026, there were at least three cases (two of which concerned the citation of fictitious authorities generated by artificial intelligence (“AI”)) in which this very issue arose. In all three cases, personal costs orders were made against the respective lawyers for their failure to properly conduct proceedings. Although these cases are non-disciplinary in nature, they indicate the Judiciary’s very real concern and provide a platform for an analysis of the ethical considerations which arise from the use of AI.
B
Objective test of reasonableness applies to presentation of authorities 2
Prior to considering the case law, it is appropriate to summarise the applicable principles and rules governing the presentation of authorities to the court. The lawyer is duty-bound “to assist in the administration of justice”, “act honourably in the interests of the administration of justice”, to “uphold the integrity of the court”
and to
“contribute to the attainment of justice” (see rr 9(1)(a) and 9(1)(b) of the Legal Profession (Professional
Conduct) Rules 2015 (“LP(PC)R”). Pursuant to these principles, he must not “knowingly mislead or attempt to mislead” the court or present information to the court which he
“knows to be false” (see rr 9(2)(a) and 9(2)(c) of the LP(PC)R).
3
These rules indicate that express intention or knowledge of falsification are conditions for accountability. However, the latter part of r 9(2)(f) of the LP(PC)R confirms that a lawyer acts unethically if
(among other expressed wrongdoings in this paragraph) he “advance[s] any submission, opinion or proposition which
[he] knows or ought reasonably to know is contrary to the law” [emphasis added].
This rule, which imposes the objective test of a reasonable lawyer, holds the lawyer accountable if he has carelessly or thoughtlessly presented a false legal submission and thereby failed to act as a reasonable lawyer.
4
This position is now confirmed by the case law. In Goh Chin Cheng v Choco Up
SG Pte Ltd [2026] SGHCR 13 (“Goh Chin
Cheng”) at [93(b)], the court cited r 9(2)(f)
of the LP(PC)R and ruled that “the propositions, holdings and quotations that
[the lawyer] claimed in his [s]ubmissions to have derived from [certain cases] were obviously
–
and very materially
–
inaccurate”. Although the court accepted that “this might have been the result of inadvertence and carelessness on [the lawyer’s] part, and not due to any intention to mislead the court, that did not change the fact that [the lawyer] ought reasonably to have known that he was misciting these authorities” [emphasis in original]. Consequently, the lawyer was ordered to pay costs personally (at [91]).
C
Using AI-generated information responsibly 5
It is appropriate to consider the regulatory infrastructure before analysing the two recent cases in which lawyers have been held accountable for misusing
AI-generated information. Apart from the substantive rules concerning the accurate presentation of legal authorities to the court (addressed in paras B2 to B3 above),
Supreme Court Registrar’s Circular No 1 of 20241, State Courts’ Registrar’s Circular
No 9 of 20242 and the Family Justice
Courts’ Registrar’s Circular No 1 of 20243
(“Registrar’s Circular”), apply specifically to the use of generative AI in court. It is 1
Supreme Court Registrar’s Circular No 1 of 2024, “Guide On the Use of Generative
Artificial
Tools
By
Court
Users”
<https://
www.judiciary.gov.sg/docs/default-source/
circulars/2024/
registrar's_circular_no_1_2024_supreme_court.
pdf> (accessed 4 June 2026).

3
entitled “Guide on the Use of Generative
Artificial Intelligence Tools by Court Users”.
6
Additionally, the Ministry of Law introduced its comprehensive “Guide for
Using Generative AI in the Legal Sector”4
on 6 March 2026 (“Ministry of Law Guide”).
Although the Ministry of Law Guide is “non-binding” (para 4), it contains detailed and helpful information which must be considered together with the Registrar’s
Circular. The Ministry of Law Guide is summarised in paras G18 to G19 below. A key concern is AI hallucination: “Broadly, hallucination refers to output that is incorrect, which can manifest in different forms, primarily through factual inaccuracy, lack of grounding or incompleteness.” (see Ministry of Law

Guide at para 16, fn 11).
7
In Tan Hai Peng Micheal v Tan Cheong
Joo [2026] SGHC 49 (“Tan Hai Peng”), the
General
Division of the
High
Court
(“General
Division”)
considered the impact of the misuse of AI on the determination of costs.5 The court found that counsel for the defendants (“Lawyer
A”) cited two “fictitious authorities” in his closing submissions which “were likely to have been generated by an AI tool” (at [6]
and [50]). Lawyer A revealed to the court that he had engaged another lawyer
(“Lawyer B”) to carry out research for the case and draft the closing submissions.
Lawyer B (who was not on the record)
claimed (in a letter to the court) that he was unaware that AI had been engaged and stated that the initial research and draft of the closing submissions were prepared by a paralegal (unnamed) who was now uncontactable (at [13]–[17]).
8
As counsel on record (“the court user”) who had the ultimate responsibility for submitting genuine authorities, Lawyer
A was obliged to ensure that any generative AI output referred to in his submissions was “independently verified, accurate, true, and appropriate” (see
Registrar’s Circular at paras 3(2)(a), 3(3),
5(1)(b) and 5(3)). The fact that Lawyer A was not aware that his closing submissions contained such output was no excuse. As the court put it (at [44]): “he had a non-delegable duty to ensure that the submissions were complete, accurate, and properly verified. By appending his signature to the final version of the submissions that were filed and served, he thereby accepted personal responsibility for all material submitted to the court.”
9
As for Lawyer B, “he ought to have recognised that a paralegal is not qualified to prepare written submissions independently, and that proper supervision and the verification of all authorities was his ultimate responsibility”
(at [46]). Lawyer B’s “duty as a solicitor and officer of the court was to ensure that every authority referred to and relied upon in the closing submissions that he was tasked to prepare was verified to actually exist and confirmed to be accurate”
[emphasis in original] (at [47]).
10
The court found it “extraordinary that between [Lawyer A] and [Lawyer B], the issue fell between two stools and neither solicitor took the time or trouble of actually verifying the existence and veracity of the
Fictitious Authorities” [emphasis in original]
(at [48]). Consequently, both lawyers were ordered to pay costs to the opposing party
(at [66]). Furthermore, having regard to the seriousness of this matter, Lawyer A was directed to inform his clients of this judgment (and provide a copy of it to the latter) and to specifically mention the personal costs orders made against him and Lawyer B (at [69]). Both Lawyer A and 2
State Courts Registrar’s Circular No 9 of 2024,
“Guide On the Use of Generative Artificial Tools
By Court Users” <https://www.judiciary.gov.sg/
docs/default-source/circulars/2024/
registrar's_circular_no_9_2024_state_courts.pdf
> (accessed 4 June 2026).
3
Family Justice Courts Registrar’s Circular No 1
of 2024, “Guide On the Use of Generative
Artificial
Tools
By
Court
Users”
<https://
www.judiciary.gov.sg/docs/default-source/
circulars/2024/
registrar's_circular_no_1_2024_family_justice_c ourts.pdf> (accessed 4 June 2026).
4
“Guide for Using Generative AI in the Legal
Sector” 2026 Ministry of Law (6 March 2026)
<https://www.mlaw.gov.sg/files/
Guide_for_using_Generative_AI_in_the_Legal_
Sector__Published_on_6_Mar_2026_.pdf>
(accessed 4 June 2026).
5
Pursuant to an earlier judgment of the court in
Tan Hai Peng Micheal v Tan Cheong Joo [2025]
SGHC 217.

4
Lawyer B were directed not to charge the clients for any time, costs or resources expended in dealing with the issue of the fictitious cases and prohibited from charging or passing on any part of the personal costs ordered to be personally paid by them (at [70]).6 11
Tan Hai Peng involved a trial of claims across three separate suits. In the prior case of Tajudin bin Gulam Rasul v Suriaya bte Haja Mohideen [2025] 5 SLR 518, the court was concerned with an application to set aside a default judgment. The claimants initially advanced a legal argument in their submissions based on a fictitious authority which had been generated by artificial intelligence (at [5]–
[10] and [34])). The defendants’ lawyer questioned the authority and counsel for the claimants (“Lawyer C”) subsequently filed amended written submissions and a supplementary bundle of authorities without the court’s permission. Pursuant to the court’s enquiry, Lawyer C initially claimed that there were “typographical errors” and “clerical errors” in the original submissions (at [15]–[17]).
12
After further questioning by the court,
Lawyer C stated that the original work had been done by a junior lawyer (at [23]), that the junior lawyer “would have run this through an AI app” (at [34]), and took responsibility by admitting that he (“Lawyer
C”) “should have checked” (at [35]). The court ordered Lawyer C to personally pay the defendant the costs of and incidental to his citation of the fictitious authority in the submissions (at [99]). It also directed both counsel to provide their respective clients with a copy of the court’s directions
“to assure the parties that the court takes a serious view of such improper conduct by advocates and solicitors” (at [100]).
13
In reaching this conclusion, the court expressed several concerns, the gravest of which was that Lawyer C “was less than candid with the [c]ourt and sought to downplay the gravity of his improper conduct” (at [82(c)]). In particular, he initially characterised the use of the fictitious case as a mere “clerical” or
“typographical” error and stated that the fictitious authority was
“wrongly”
or
“inadvertently” cited (at [82(c)(i)]). He failed to voluntarily disclose that the fictitious authority was generated by a generative AI tool until he was questioned by the court (at [82(c)(ii)]). He described the fictitious authority as a “placeholder”
case, which “trivialised the seriousness of his improper conduct and demonstrated a continued failure to appreciate its gravity”
(at [82(c)(iii)]).
D
What is at stake 14
An officer of the court has a fundamental duty to ensure that the authorities he presents to the court are accurate (see para B2 above). If he or any person under his supervision uses generative AI, the lawyer conducting the case is personally responsible for making certain that that the authorities he puts before the court are accurate, true and appropriate (see Registrar’s Circular at paras 3(2)(a), 3(3), 5(1) and 5(3); see also para C8 above). In particular, he is obliged to refer to the checklist in para 5(3), which sets out all the steps that must be taken to ensure accuracy. Indeed, the lawyer must be “prepared to identify the specific portions of the Court Documents which used AI-generated content, and explain to the Court how [he has] verified the output produced by a Generative AI tool” (see
Registrar’s Circular at para 5(4)).
15
A failure to abide fully with the
Registrar’s Circular may result in multiple breaches of the LP(PC)R. Apart from the lawyer’s breach of duty to the court, inaccuracy in the presentation of authorities or a fictitious case may result in a miscarriage of justice if the cited authority affects the decision. Moreover, such conduct would amount to a breach of his duty to uphold the standing and integrity of the administration of justice and the legal profession and adversely affect the public image of the profession (see r 4(b) of the LP(PC)R). His responsibility to act professionally and fairly towards the opposing party (see rr 8(1) and 8(3) of the 6
See also Goh Chin Cheng v Choco Up SG Pte
Ltd [2026] SGHCR 13 at [100].

5
LP(PC)R) and opposing counsel (see rr 7(1)
and 7(2) of the LP(PC)R) would also be compromised. Regarding his client, lack of diligence and competence in using generative AI may compromise the case or result in complications (this may amount to a breach of rr 5(1) and 5(2)(c) of the
LP(PC)R). As the cases show, a lawyer may not excuse himself for misuse of generative
AI by arguing that some other person carried out the research. Rule 32 of the
LP(PC)R states: “A legal practitioner must, regardless of the legal practitioner’s designation in a law practice, exercise proper supervision over the staff working under the legal practitioner in the law practice.”
E
Responsibility of law practices 16
The management of a law practice must be cognisant of its responsibility under r 35(7) of the LP(PC)R) to “ensure that the law practice complies with the requirements of the [Legal Profession] Act
(including [the LP(PC)R])”. Although the
LP(PC)R does not expressly refer to the use of generative AI, a breach of the
Registrar’s Circular or contravention of a relevant case law ruling concerning this area could constitute a breach of any of rules mentioned in the preceding paragraph. Rule 35(4) requires the law practice to have “in place adequate systems, policies and controls” for ensuring that legislation and regulations relating to client’s money, conflicts of interests and client confidentiality are complied with.
Generative AI raises the concern that client confidentiality may be compromised, particularly when client information is processed through third-party systems (see the Ministry of Law

Guide at paras 21–24 (also referred to in para C6 and G18–G19 of text); see also
Annex A of the Ministry of Law Guide for examples of steps which may be taken to ensure confidentiality when using generative AI tools).
F
Sanctions 17
A failure to properly use generative AI may be grounds for striking off, suspension or other sanctions if the lawyer is found
“guilty of fraudulent or grossly improper conduct” or to have breached a rule of the LP(PC)R (see ss 83(2)(b)(i) and 83A(2)(i)
of the Legal Profession Act 1966 (2020 Rev
Ed)) or is “guilty of such misconduct unbefitting an advocate and solicitor as an officer of the Supreme Court or as a member of an honourable profession” (see ss 83(2)(h) and 83A(2)(h) of that Act). In
Tan Hai Peng (see paras C7–C10 above), the General Division observed (at [84]) that misconduct in the use of generative AI
“may, if an appropriate case arises in the future, result in the errant solicitor(s) being subjected to disciplinary action”.
G
Ministry of Law’s Guide for using generative AI in the legal sector 18
As mentioned in para C6 above, the recently introduced Ministry of Law Guide provides comprehensive guidelines for the use of generative AI in the legal sector.
Although it is not binding, it is a crucial document which explains the issues and assists law practices and lawyers in navigating this domain, setting up proper systems of control, and engaging AI tools.
A summary of the guide is appropriate here.
The purpose, scope and applicability, definition and concepts are set out in Part 1. Part 2 concerns the evolution of AI in legal practice and the rise of generative AI. The key principles which apply to legal work are expressed in
Part 3. They concern professional ethics, confidentiality and transparency. Practical examples of how law practices and in-house legal teams have applied these principles are provided in Annex A.
19
The implementation of generative AI in legal practice is the subject-matter of
Part 4. The five sequential steps proposed concern the development of an AI adoption framework, diagnosing and analysing needs, identifying and evaluating generative
AI tools, implementation and training, and

6
continuous review and improvement.
Annex B includes illustrations of how generative AI has been implemented in legal practice. Annex C sets out sample templates and clauses for generative AI usage policy. Annex D offers a sample evaluation checklist. Annex E provides a checklist for cyber security and data protection, data handling measures and available support and training resources.
Although the Ministry of Law Guide is not legally binding, lawyers and law practices should be aware that the failure to give it proper attention may result in avoidable breaches of multiple ethical rules (see paras D14–D15 and E16 above).

Published by Academy Publishing
LawNet | Singapore Law Watch
DISCLAIMER
Views expressed by the author are not necessarily those of Academy Publishing nor SAL. Whilst every effort has been made to ensure that the information contained in this work is correct, the author, Academy Publishing and SAL disclaim all liability and responsibility for any error or omission in this publication, and in respect of anything, or the consequences of anything, done or omitted to be done by any person in reliance, whether wholly or partially, upon the whole or any part of the contents of this publication.
COPYRIGHT
© 2026 Singapore Academy of Law.
All rights reserved. No part of this publication may be reproduced, stored in any retrieval system, or transmitted, in any form or by any means, whether electronic or mechanical, including photocopying and recording, without the written permission of the copyright holder. All enquiries seeking such permission should be addressed to:
Singapore Academy of Law 1 Coleman Street #08-06 The Adelphi Singapore 179803 Tel No: (+65)
6332 4388 Fax No: (+65) 6333 9747 E-mail: academypublishing@sal.org.sg

What's the problem? (required)

Source link (optional)

Describe the problem (and edit the text or pick an issue type) to submit.