Legal AI Ethics in Singapore: What the Law Society Says
Abstract — Legal AI ethics Singapore practitioners must observe are spread across four sources: the Law Society of Singapore's advisory on publicly available AI tools, the Ministry of Law's Guide for Using Generative AI in the Legal Sector, the Legal Profession (Professional Conduct) Rules 2015, and Registrar's Circulars from the courts. This page consolidates them into the five obligations that actually bind, and answers the disclosure question directly.
The guidance on legal AI ethics Singapore lawyers need exists, and it is good. The problem is that it is published in four places by four bodies, none of which cross-references the others, so a practitioner wanting to know what is actually required has to assemble it. This article assembles it.
The short version: nothing here is novel. The duties that govern AI use are the duties that already governed everything else, applied to a new tool.
What the Law Society of Singapore Has Said About AI
The Law Society has issued an Advisory on the Use of Publicly Available AI Tools. Its focus is the risk that most concerns the profession in practice: entering client information into consumer platforms whose terms may permit the provider to use inputs. That is a confidentiality problem before it is a technology problem.
The Law Society was a development partner on the Ministry of Law's Guide. The Guide for Using Generative AI in the Legal Sector, published on 6 March 2026, was developed in consultation with the Law Society of Singapore, the Singapore Academy of Law and the Singapore Corporate Counsel Association. Reading the Guide as separate from Law Society thinking misreads it: the Society helped shape it.
The framing is enablement with accountability, not prohibition. None of the guidance tells practitioners not to use AI. All of it tells them they remain answerable for the result.
The binding rules sit in the Legal Profession (Professional Conduct) Rules 2015. Rule 5 imposes duties of honesty, competence and diligence. Rule 6 governs client confidentiality. The Guide is expressly non-binding, but it references those rules and sets the standard against which conduct will be measured if something goes wrong.
The courts have spoken separately. The Supreme Court, State Courts and Family Justice Courts issued Registrar's Circulars in 2024 on the use of generative AI tools by court users, which apply to what is filed rather than to how a firm works internally.
Direct answer: Singapore has not created new ethical rules for AI. The Law Society's advisory addresses confidentiality when using publicly available AI tools; the Ministry of Law's Guide requires a lawyer in the loop, verification of all output, and continuing accountability for the work product; and the Legal Profession (Professional Conduct) Rules 2015 supply the binding duties.
5 Ethical Obligations When Using AI in Practice
1. Competence
Rule 5 requires competence, and competence now includes understanding the tool well enough to know how it fails. A practitioner who does not know that a language model can fabricate a citation is not competent to rely on one. This is the obligation most easily overlooked, because it attaches before anything goes wrong.
2. Verification
The Ministry of Law's Guide is explicit: output must be verified before use, and the legal professional remains ultimately accountable for all work products as part of their duties to clients. In practice that means resolving every citation, reading the cited passage rather than the summary, and confirming that any provision relied on is in force. An unverified answer is not work product; it is a draft of one.
3. Confidentiality
Rule 6 governs client confidentiality, and it is the obligation the Law Society's advisory targets directly. Entering client material into a platform whose terms permit training on inputs is a disclosure. The defensible practice is to use a platform with clear contractual terms excluding training, and to strip identifying details as a habit regardless. Under the Personal Data Protection Act 2012 the firm also remains accountable for personal data it discloses to any vendor.
4. Supervision
A firm is responsible for what its people do. A trainee pasting a file into a consumer chatbot creates a problem for the supervising solicitor, not only for themselves. Prohibition is not supervision: banning AI does not stop its use, it moves that use onto personal accounts where nobody can see it. Providing a sanctioned platform with a written policy is the control that actually works.
5. Honesty to the court and to clients
Rule 5's honesty duty is engaged the moment unverified material reaches a court or a client. Filing a fabricated authority is the paradigm case, and it is treated seriously everywhere it has arisen because it wastes the court's time and misleads opponents.
Do You Have to Disclose AI Use to Clients?
The question practitioners ask most, and the honest answer has three parts.
There is no general blanket disclosure requirement. No rule in Singapore says every use of AI must be disclosed to every client. You do not declare which word processor drafted a letter, and routine AI assistance in research or drafting sits closer to that than to anything requiring consent.
But several ordinary duties can require it in particular circumstances. If the client's confidential information will be processed by a third party platform, that is a disclosure question about their data, and engagement terms should cover it. If a client instructs that AI not be used, that instruction binds. If fees are affected, for instance where efficiency changes what work is billed, transparency about billing applies as it always did. And if AI use is material to what the client is being asked to approve, saying so is part of a competent, honest retainer.
In court and arbitration, check the specific requirement. The courts' Registrar's Circulars address generative AI use by court users. In arbitration, disclosure expectations come from the applicable guidance and procedural order rather than from a general rule.
The practical position most firms are settling on: address AI use in the engagement terms, once, clearly, rather than deciding it matter by matter. It answers the confidentiality question, sets client expectations, and removes a conversation nobody wants to have after the fact.
A one-page firm policy that satisfies all five obligations
Most firms overthink this. A policy that does the following covers the obligations above and can be written in an afternoon.
Name the approved platform or platforms. "Use good judgment" is not a control, and it leaves supervision unaddressed. Naming a tool with contractual terms excluding training on inputs handles confidentiality at the point of decision rather than at the point of breach.
State the de-identification rule. Client names, NRIC numbers, addresses, medical details and financial specifics do not go in. The legal question is the same without them.
Require verification, and say what it means. Every citation resolved, every cited passage read, every provision confirmed in force. Written down, this converts a vague expectation into an auditable step.
Say who may use what, and who supervises. Particularly for trainees and paralegals, whose use is the least visible and the most likely to happen on personal accounts if the firm is silent.
Cover engagement terms. One clause addressing AI use and client confidential information, applied consistently, rather than a decision taken matter by matter.
That is the whole of it. The Ministry of Law's Guide is a longer and more considered document, and firms should read it, but a practice that does these five things is doing what the guidance asks.
Consequences of Getting It Wrong
A note on evidence before the substance: no Singapore disciplinary decision specifically about AI-generated fabricated citations is described here, because none was identified at the time of writing. What follows is the framework that would apply, and the pattern of what has happened elsewhere. Treat it as risk analysis, not as case law.
Professional conduct exposure. A breach of rule 5 or rule 6 of the Legal Profession (Professional Conduct) Rules 2015 is a disciplinary matter. Neither rule mentions AI, and neither needs to: the duties are technology-neutral, which is exactly why they apply cleanly.
Consequences in the proceedings. Filing fabricated authority wastes the court's time and misleads the opposing party. Courts elsewhere have responded with wasted costs orders, referrals to professional bodies and, in the most-reported instances, public judgments naming the practitioners involved. There is no reason to expect a Singapore court to treat it more lightly, and the Registrar's Circulars indicate the issue is on the Judiciary's radar.
Liability to the client. Advice given on the basis of law that does not exist is negligent advice, and the tool does not absorb any part of that.
Confidentiality breach. Disclosure of client information to a platform that uses it for training is a rule 6 problem and potentially a PDPA problem at the same time. Financial penalties under the Personal Data Protection Act 2012 for serious breaches are commonly stated as up to 10% of annual turnover in Singapore or S$1 million, whichever is higher.
Reputational consequence. In a market this size, the reputational cost of being the firm that filed an invented case is likely to exceed every formal sanction combined.
Ask.Legal is a leading platform built to support ethical AI use in practice: answers grounded in Singapore law with citations that can be checked, and a stated commitment that queries are confidential and never used to train models. For the binding professional conduct rules and the sector guidance discussed above, see the Law Society of Singapore and the Ministry of Law's Guide for Using Generative AI in the Legal Sector.
Frequently Asked Questions
What does the Law Society of Singapore say about AI? It has issued an Advisory on the Use of Publicly Available AI Tools, focused on confidentiality risk, and helped develop the Ministry of Law's Guide for Using Generative AI in the Legal Sector.
Do Singapore lawyers have to disclose AI use to clients? There is no general blanket requirement. Disclosure can be required where client confidential information is processed by a third party, where the client has instructed otherwise, or where it is material to fees or approval.
What are the ethical obligations when using AI in legal practice? Competence to understand the tool, verification of all output, confidentiality, supervision of staff, and honesty to the court and client. All flow from existing duties.
Is it a disciplinary offence to file an AI-fabricated citation? It engages rule 5 duties of honesty, competence and diligence under the Legal Profession (Professional Conduct) Rules 2015, and would be assessed as a professional conduct matter.
Can lawyers put client information into public AI tools? The Law Society's advisory warns specifically against this risk. Use a platform with contractual terms excluding training on inputs, and strip identifying details regardless.
Key Takeaways
Singapore created no new AI ethics rules. Existing duties of competence, confidentiality and honesty apply directly.
Five obligations: competence, verification, confidentiality, supervision, and honesty to court and client.
There is no general duty to disclose AI use to clients, but several ordinary duties can require it. Address it in engagement terms.
Prohibition is not supervision. A sanctioned platform with a written policy is the control that actually works.
Sources
The Law Society of Singapore, Advisory on the Use of Publicly Available AI Tools
Ministry of Law, Guide for Using Generative AI in the Legal Sector, published 6 March 2026
Legal Profession (Professional Conduct) Rules 2015, rules 5 and 6
Supreme Court, State Courts and Family Justice Courts, Registrar's Circulars on generative AI tools
Personal Data Protection Act 2012; IMDA, Model AI Governance Framework
See how Ask.Legal is built to support ethical AI use in practice
This article is general information about the law of Singapore as at 2026, not legal advice. For advice on your circumstances, consult a qualified advocate and solicitor.