A waste of precious judicial time and resources at best, and a potential miscarriage of justice at worst. These were the words of Justice S Mohan, who ordered two lawyers to each pay $5,000 in personal costs for citing fictitious authorities that were likely generated by AI in court.
Singapore Lawyers Cite Fake Cases
The civil case before Justice Mohan involved the estate of a deceased individual. The claims were brought by the sons of the deceased, who were also co-executors of his estate. The defendants comprised four brothers and Fong Tat Holding Co Pte Ltd, a company in which the brothers were directors and shareholders. The claims concerned several loans made by the deceased to the defendants. The problem arose when lawyers for the defendants filed closing submissions citing two nonexistent authorities.
Making reference to the Singapore judiciary’s AI Circular, Justice Mohan reminded lawyers that they must be personally responsible for ensuring that “all matters placed before the court”, including case authorities, are verified to be true and accurate. He then rejected the duo’s request to resolve the matter of costs privately. Justice Mohan reasoned that their actions risked contravening the AI Circular, Supreme Court Practice Directions, and the Legal Profession (Professional Conduct) Rules 2015. He went on to state, “It is for this reason that I cannot countenance Mr. Goh and Mr. Sidhu’s requests that they be allowed to resolve the question of costs privately among the parties. It is in the interests of both the public and the legal profession that the matter not be dealt with privately.”
Unfortunately, Mr. Goh and Mr. Sidhu were not the only ones caught citing fake cases in court.
In a hearing on 22 July 2025, Assistant Registrar Tan Yu Qing ordered Mr. Lalwani to pay $800 in personal costs for citing a “fictitious legal authority” in his Written Submissions. This was Singapore’s first reported case of phantom, or AI-generated authorities in court. Though Mr. Lalwani attributed the error to the work of a junior lawyer who had been called to the bar not long ago, AR Tan maintained that court users should be “fully responsible for the content in all their court documents”—including AI-generated references.
AI-generated case cited by litigant-in-person
A man’s application for a personal protection order (PPO) was dismissed after Magistrate Soh Kian Peng found his claims of “emotional and psychological abuse” unconvincing. The man had also used AI tools to assist in the preparation of statements submitted to court.
Although the man claimed to have only deployed AI tools to create first drafts of documents and to generate templates, Magistrate Soh noted that he had cited a fictitious case. He also noted how the man would shift between referring to himself in the first and third person in court documents.
Magistrate Soh then urged court users who use generative AI to take responsibility for all outputs and not give into the temptation to offload their thinking to AI tools.
AI’s Impact on the Legal Profession
On 11 September 2025, the Infocomm Media Development Authority (IMDA) and the Singapore Academy of Law (SAL) announced the launch of LawNet 4.0, which is powered by the latest release of the GPT-Legal Q&A model. The IMDA and SAL estimate that more than 75% of lawyers in Singapore will benefit from a shorter legal research journey as a result of this update.
The deployment of AI extends beyond LawNet; the legal profession is feeling the effects of delegating work typically done by junior lawyers and fresh graduates to AI. A Channel News Asia article on the subject was published on 20 March 2026 and featured several young lawyers and established law dons. Melvin Loh, who is a senior lecturer of law at SUSS, reveals that one of the biggest challenges he faces as an educator is students who turn to generative AI before building a strong foundation of domain knowledge. To mitigate the dulling of “thinking and reasoning skills”, the NUS Faculty of Law will mandate the return of closed-book law exams in the next academic year. Chief Strategy Officer Professor Tan Cheng Han hopes this will help students think critically about the law, as well as understand and apply legal principles and doctrines even as aspects of legal research are becoming increasingly automated.
When it comes to artificial intelligence and the law, outgoing Chief Justice Sundaresh Menon is a firm proponent of the “traffic light model”. Red zones prohibit the use of any AI tools; amber zones are for areas where AI use can be considered, while green zones are where AI tools can be freely deployed.
Traffic light model of AI use in courts
| Red | No AI use should be deployed where an exercise of judgement is involved |
| Amber | Collating of arguments, drafting arguments for laypeople, and summarisation of submissions may be areas where AI tools can help |
| Green | Summarisation of documents and translations are deemed to be areas where AI can be safely deployed |
The Chief Justice emphasised that judgements cannot be outsourced to AI tools, as people who stand before the law would want to be judged by people who understand their life, their emotions and their situation. All things that cannot be measured or weighed by large language models or machines.
AI and legal liability
In a 2024 NUS Law Research Blog article co-authored by Associate Professor Daniel Seng Kiat Boon and Professor Tan Cheng Han (now Law Society President) on Agency Law and Artificial Intelligence, the two authors held that AI tools or agents cannot be held liable for wrongs unless granted personhood in the same vein as corporations. Calling it the “instrumentality principle”, the two authors go on to say that such tools were only instruments of legal entities or persons, not a legal agent.
Infocomm Media Development Authority’s (IMDA) Legal Responsibility for AI Agents discussion paper released in May 2026 reinforces this line of thinking, stating that both individuals and legal entities are accountable for their actions, regardless of whether they use AI tools to make decisions or take actions on their behalf.
The issue of how to assign blame when non-deterministic AI systems are involved is more complex. IMDA’s paper explored the feasibility of negligence and strict liability regimes, with the view that all participating actors may have to share some degree of the risks.
In contrast, the European Law Institute’s (ELI) Guiding Principles for Automated Decision-Making in the EU released in 2022 assigns blame squarely to the operator. In the ELI’s view, the operator is best positioned to assess and manage risks, as it is “in control of the risks connected with the [AI system] and benefits from its operation”. In the ELI’s hypothetical scenario, a university implementing an algorithm-driven mail system which delivers parcels by drone would be directly responsible for any injuries or property damage that ensues.
Can AI lawyer assistants provide value to clients?
With the National AI Strategy in full swing, it is unsurprising that AI lawyer assistants and legal consultants are appearing on the scene. While some law firms have already begun using mass market AI tools such as Microsoft CoPilot to streamline their workflow, other firms prefer specialised ones such as Harvey and GPT-Legal. These tools are used to aid in the completion of onerous tasks such as the summarisation of lengthy documents, as well as the drafting and reviewing of legal filings, among other uses.
So far, however, it remains unclear whether AI lawyer assistants are benefiting clients even as billable time is ostensibly reduced.
Artificial Intelligence in Court: Phantom Cases Go Worldwide
Singapore is far from being the only country in the world where lawyers are using AI tools in court.
June 2026
- Two Munich-based publishing companies took Google to court over its AI overview feature and concerns of reputational damage. The AI-generated summary, which appears on top of the search engine results page, had falsely linked the two companies to dubious business practices. In a landmark ruling, the Munich court determined that Google should be held responsible for spreading false claims and ordered the search engine operator to pay 80% of legal costs. Though Google has said that it will appeal the court’s ruling, the court’s decision nevertheless challenges the presumption that AI tools can never be held to account.
- South Korea’s judiciary is mulling over the imposition of fines for lawyers who cite “ghost cases” in court and disclosure rules connected to AI tools usage. This comes after a rise in lawyers submitting false or irrelevant citations.
July 2026
- A federal judge calls out a Department of Justice (DOJ) lawyer for citing a fake AI case. Chief Judge Hala Y. Jarbou issued a warning after noting that a reference to a nonexistent US Court of Appeals for the Sixth Circuit had been made. In April, a similar case of a DOJ lawyer being taken to task for an AI-generated brief filled with fabrications was reported.
Can Lawyers Use AI Tools in Singapore Courts?
Yes, lawyers are permitted to use AI tools in Singapore courts.
The Singapore judiciary’s AI Circular explicitly states that, “The Court does not prohibit the use of Generative AI tools to prepare Court Documents”. However, the Circular also makes it clear that lawyers are responsible for ensuring the accuracy of all AI-generated output used in Court Documents. Legal professionals should ensure that all such information is:
- Accurate
- Relevant
- Does not infringe on intellectual property rights
It is clear that leading law authorities in Singapore believe that AI tools can be harnessed for good in a legal setting, if used responsibly.
Chief Justice Menon himself has said that AI tools are here to stay; he is also co-leading a committee with Law Minister Edwin Tong to help the legal profession meet multifaceted challenges. According to the Chief Justice, the future of the legal profession depends on whether lawyers can find a way to work with technology to “multiply the potential of [their] legal skills and knowledge”. Meanwhile, Professor Tan Cheng Han has highlighted the “practical and meaningful adoption of AI by smaller firms” as one of his immediate priorities in his new role as Law Society President.
