In the concluding article of our three-part series, "Reinventing Law: Singapore's Legal Profession in Transformation", we explored how artificial intelligence, digitalisation and evolving client expectations are likely to reshape the legal profession by 2030. We observed that generative AI is rapidly moving from experimentation to becoming an integral part of legal practice, transforming how lawyers conduct research, review documents, manage transactions and deliver legal services.
Yet amid the excitement surrounding AI, an important question remains: how can the profession embrace innovation without compromising the ethical standards that underpin public trust?
This question is not unique to lawyers. It is one that every profession built on expertise and public confidence must confront. Auditors, for example, have long leveraged sophisticated data analytics, automation, and technology-enabled audit procedures. However, despite these advancements, professional standards have never allowed technology to replace professional judgement, scepticism, or accountability. The same principle applies to the legal profession.
Artificial intelligence may assist the lawyer, but it cannot replace the lawyer's professional responsibility.
As the legal profession embraces artificial intelligence, maintaining public trust will depend not only on how quickly firms adopt innovative technologies, but also on how effectively they govern their use. The challenge facing the profession is therefore not whether to use AI, but how to do so responsibly, ethically and in a manner consistent with the core principles that underpin the administration of justice.
The Emerging Ethics Challenge
One of the most widely discussed risks associated with generative AI is its tendency to generate information that appears credible but is factually incorrect, commonly referred to as "hallucination".
While this may appear to be a technological limitation, recent incidents demonstrate that the real risk lies in how professionals use the technology.
The most well-known example was the 2023 case of Mata v Avianca Inc.1 in the United States, where lawyers relied on ChatGPT-generated legal research that cited six entirely fictitious cases. The court subsequently imposed sanctions of US$5,000, emphasising that lawyers remain responsible for verifying the accuracy of all authorities cited in their submissions.
More recently, Singapore has encountered similar issues. In Tajudin bin Gulam Rasul and another v Suriaya bte Haja Mohideen [2025] SGHCR 332 , the court dealt directly with a lawyer who cited a fictitious authority generated by a generative AI application. The court reaffirmed that lawyers have a fundamental duty to independently verify all materials placed before the court and described the citation of fictitious AI-generated authorities as "wholly impermissible".
These cases are often portrayed as failures of artificial intelligence. In reality, they are failures of professional scepticism and ethical judgement.
The technology generated the content, but the professionals failed to perform the verification procedures that would ordinarily be expected before relying on any source of information. From a governance perspective, the root cause lies not in the technology itself, but in the breakdown of review, supervision, and quality control processes.
Lessons from the Audit Profession
The legal profession may find useful parallels in the audit and assurance profession.
Auditors routinely use sophisticated technologies, data analytics and increasingly AI to analyse large datasets and improve audit effectiveness. However, no auditing standard permits an auditor to accept an output simply because it was generated by a sophisticated tool.
Professional scepticism remains a cornerstone of the audit process.
Audit evidence must be evaluated, corroborated and challenged before conclusions are reached. Responsibility for the audit opinion rests with the engagement partner, regardless of the technologies used during the engagement.
The same principle should apply to legal practice.
AI may assist with legal research, contract review, due diligence, and drafting. It may improve efficiency and enhance productivity. However, responsibility for the accuracy of legal authorities, the quality of legal advice and the protection of client interests remains with the lawyer.
In this regard, AI should be viewed as an exceptionally capable assistant, but never as an expert. It can accelerate processes and support decision-making, but it cannot assume professional responsibility.
What Can Be Done Better?
The question facing law firms is therefore not whether AI should be used, but how its use should be governed.
As AI becomes embedded within legal workflows, firms should consider implementing governance frameworks similar to those adopted for other high-risk technologies and professional processes.
1. Establish Clear AI Governance Policies
The most effective organisations do not leave technology adoption to individual discretion. Instead, they establish clear policies governing:
- Approved AI tools and platforms
- Permitted use cases
- Confidentiality and data protection requirements
- Client disclosure obligations
- Review and approval procedures
Just as firms have policies governing cybersecurity, conflicts of interest and quality management, AI governance should become part of the firm's overall risk management framework.
2. Implement Mandatory Verification Procedures
A simple principle should apply: If a lawyer cannot independently verify an AI-generated output, it should not be relied upon.
This principle mirrors a fundamental concept familiar to auditors: conclusions should only be based on evidence that is reliable, relevant, and capable of independent verification.
AI-generated content should never bypass established review procedures simply because it appears authoritative.
3. Strengthen Supervision and Accountability
Many of the AI-related failures reported globally have not arisen because AI systems malfunctioned. Rather, they occurred because insufficient supervision allowed erroneous outputs to pass through quality control processes undetected.
As AI becomes more pervasive, firms will need to revisit existing review mechanisms and clarify accountability for AI-assisted work products.
4. Invest in Ethical AI Competency
Technical proficiency alone will not be sufficient.
Lawyers will need to understand:
- AI limitations and hallucination risks
- Bias and explainability concerns
- Confidentiality implications
- Regulatory expectations regarding AI use
Recognising this, Singapore's Ministry of Law launched the Guide for Using Generative Artificial Intelligence in the Legal Sector3 on 6 March 2026, signalling that responsible AI use is becoming a vital component of professional practice.
Looking Ahead
Artificial intelligence will undoubtedly become a permanent feature of legal practice. The profession's future will not be defined by whether lawyers use AI, but by how responsibly they use it.
The fundamental principles underpinning the legal profession remain unchanged:
- Competence
- Integrity
- Objectivity
- Confidentiality and
- Professional behaviour.
Significantly, these are not principles unique to lawyers. They are equally foundational to other trusted professions, including auditing and assurance, where public confidence depends on the consistent exercise of professional judgement, ethical conduct, and accountability. Technology may transform how professional services are delivered, but it does not diminish the responsibilities that accompany them.
The most successful law firms of the future will not necessarily be those that adopt AI the fastest. Rather, they will be those that combine technological innovation with strong governance, rigorous quality control, and an unwavering commitment to professional ethics.
In the age of AI, the profession's greatest safeguard is not technology itself, but the judgement, scepticism and integrity of the professionals who use it.
------------------
1 MATA v. AVIANCA INC (2023) | FindLaw
2 [2025] SGHCR 33
3 Guide_for_using_Generative_AI_in_the_Legal_Sector__Published_on_6_Mar_2026_.pdf