The text
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates 1 Ms Hany Soh asked the Minister for Education in respect of the "O" Level English Language oral examination conducted on 15 July 2026, whether the Ministry intends to review and improve the way the oral examination is conducted notwithstanding the Singapore Examinations and Assessment Board's finding that there was no discrepancy in the question displayed on the candidates’ and examiners' screens, to minimise risks of students being unnecessarily affected.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates 2 Ms Hany Soh asked the Minister for Education with respect to the "O" Level English Language oral examination conducted on 15 July 2026 (a) whether any requests for assistance from students or their parents have been received by Singapore Examinations and Assessment Board (SEAB) due to the alleged discrepancy between the question displayed on the candidates’ and the examiners’ screens; and (b) if so, how has SEAB been assisting these students.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates 3 Mr Low Wu Yang Andre asked the Minister for Education before the 2026 "O" Level English results are finalised (a) whether the Singapore Examinations and Assessment Board (SEAB) will compare candidates’ scores and responses from the 15 July 2026 oral session against other sessions; and (b) what remedies will apply if this shows a material session-level or examination-centre-level anomaly despite SEAB’s finding that one version of the examination video and questions was released.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Mr Speaker, may I please address oral Question Nos 1 to 3 on today's Order Paper?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Please proceed.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates My response will also address related oral and written questions filed for the Sitting on 4 August 2026.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, as the questions are about the GCE "O" Level English Oral Language Examination on 15 July, let me first explain the examination process.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Candidates are first seated at the silent preparation station. The candidates view a computer screen with three components simultaneously displayed. The examination instructions, a text prompt and a video.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates They are given 10 minutes at this station to prepare a planned response to their oral presentation. After 10 minutes, the candidate is called to the examiners' station. The oral examiner reads out the same text prompt and invites the candidate to commence the presentation. There is another examiner present.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Candidates are given two minutes to deliver their planned response. This is followed by the spoken interaction component, where the candidate engages in a discussion with the oral examiners on a topic based on the video clip.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates All examination materials are subjected to a rigorous and secure process of authoring, review and approval. The oral examinations are accessed through the Singapore Examinations and Assessment Board's (SEAB's) e-Examination system. Digital logs monitor all activities carried out in the system to ensure the integrity of the processes.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates For each oral examination session, only one version of the examination materials is stored on SEAB's e-Examination system. All examination centres access the same materials when they log into SEAB's e-Examination system. The materials are encrypted and can only be accessed by authorised personnel.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Before the start of each session, the Assessment Specialist at SEAB checks that the prompt shown on-screen to candidates during the silent preparation and the prompt shown to the oral examiners tally exactly with the prompt on the question paper in hard copy. Once the examination has commenced, the materials cannot be amended. During the examination, all stations in the examination centres, including the candidates silent preparation stations and the oral examiners stations, access the same approved, secured, encrypted examination materials from the e-Examination system.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates There were 389 oral examination teams across schools, with each team comprising two trained oral examiners assessing 13 candidates per session. The video presented to candidates depicted street barbers at work. The prompt for the planned response segment was, "Who do you think would go to a place like this for a haircut? Why do you say so?"
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates On the day of the examination, 173 candidates from 76 different schools reported that the prompt for the planned response segment displayed on their screen at the silent preparation station was different from the one that was posed by the oral examiners.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates In the following days, the number of candidates who reported similar issues concerning the 15 July oral examination increased to 432 from 102 schools. Close to 5,000 candidates sat for the GCE "O" Level English language oral examination on 15 July 2026. On 15 July itself, SEAB took immediate actions to investigate. SEAB verified the examination materials by making system level checks and by logging into candidates' individual accounts in the e-Examination system, which shows the prompt and video accessed by the candidates. The investigation showed that the prompt on the candidates' screens match the prompt in the approved examination materials.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates SEAB subsequently also reviewed all the digital logs of the authoring and delivery of the examination material from SEAB's e-Examination system and the digital logs of what was displayed on individual candidates' screens during the silent preparation, for all candidates who reported a discrepancy. In addition, cybersecurity checks were conducted to ascertain if the e-Examination system had been compromised.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Investigations showed that in all examination centres, oral examiners and candidates accessed the same examination materials during the oral examination on 15 July 2026. There was no evidence of a system malfunction. There was no discrepancy in the prompt displayed to candidates and oral examiners. There was only one source that all candidates' and examiners' accounts accessed. There was also no cybersecurity incident detected.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates We also observed no discernible pattern of these incidents. The 432 candidates who reported this discrepancy came from 200 different oral examiner teams. There was also no particular order to the affected candidate. For instance, it was the second and eighth out of 13 candidates for one oral exam team. So, in that sequence, this oral exam team would examine 13 candidates, and it was the second and eighth candidate reporting a discrepancy with the other candidates reporting no discrepancy. And for another oral team, it was the third and 10th candidate, for example.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates To be clear, the discrepancies reported were not about what the oral examiners asked – only that the recollection of the text prompt at the preparation station was different from what the oral examiner read out.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates The reported discrepancies also demonstrated significant variation, with different phrases recalled by different students, including by students that were examined at the same centre and by the same oral examination team. The actual prompt shown was, "Who do you think would go to a place like this for a haircut? Why do you say so?" Some candidates reported reading the prompt as, "Would you go?" Others reported reading the prompt as, "Would you visit?", "Would you be interested?", "Would you like?", "Would you want?", "Would you cut your hair at the place shown?", "Would you get your hair cut?" and "Why would I go?"
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Putting together the technical details, the distribution pattern of the incidents across the stations, schools and time, as well as the variety of phrases reported by the students. SEAB has concluded the cause of these incidents was not technical in nature, nor the result of a deliberate or malicious act.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates We understand that the experience would have been unsettling for the affected candidates. Parents and candidates can be assured that our oral examiners are experienced teachers who assessed candidates based on what they demonstrated during the examination, including how they developed and presented their ideas and opinions, and how effectively they engaged the examiner during the discussion.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates The preliminary data shows that the mean score for the oral examination on 15 July is comparable to that of the other four sessions in 2026. In addition, the mean score of the 432 candidates who reported the prompt discrepancy is also comparable to the mean score of the remaining candidates who sat for the oral examination on 15 July 2026.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates So, therefore, as a group, the candidates who reported discrepancies fared as well as other candidates who did not report any discrepancy.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates There might be some candidates who experienced significant distress arising from the circumstances. This might have affected their ability to provide a suitable response. Schools are working with these candidates to apply for special consideration. Special consideration is an established evidence-based process by examination boards to review the final subject grades for candidates whose performance had been adversely affected by circumstances beyond their control during the examinations. This is to ensure fair assessment. Each application will be evaluated on its merits.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates We thank candidates, parents and schools for bringing their concerns to our attention. We take these reports very seriously. SEAB conducted a thorough investigation before coming to the conclusion that the same prompt was displayed for all candidates. SEAB will review the oral examination processes for students to read the oral prompt more carefully during the 10-minute preparation time. SEAB also remains committed to ensuring that all candidates, including those who might have misread the prompt, are assessed fairly.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Ms Hany Soh.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates I thank the Senior Minister of State for his substantive reply. In relation to the special considerations that he shared with this House earlier, I also understand from my resident that that has been the reply from SEAB to one of my residents whose child has been affected.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates In this regard, can I enquire so far in relation to the over 500 cases that were reported of discrepancies that were highlighted – 400 plus in total, I beg your pardon – have any of these students followed up with a special consideration application? And in such an instance, what will be the process taken and whether the students who had put in the appeal would eventually be updated that their results have taken in the circumstances that had transpired?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, I thank Ms Hany Soh for the question. Certainly, the applicants for special consideration will be updated and engaged on the process.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates The process of special consideration is still in process. So, I think it is perhaps not appropriate for me to comment on the specifics. Each case will be evaluated. The process of special consideration is to look at the particular circumstances of each case.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Certainly, some applications have been made and potentially further applications will be made. So, I think we have to let that process play out.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Mr Andre Low.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Thank you, Speaker. I have two supplementary questions.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates First, if I may paraphrase what I am hearing as the explanation or what the investigation has uncovered, is that the system architecture by its very nature would not permit the incident that students are describing from happening – because of the nature of the fact that it is a single question stored in the database and it is a single question transmitted to all terminals. That being said, I understand also that centralised logs were also checked to make sure that this was indeed what happened.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Can the Ministry share if the terminals themselves maintained any logs and whether these logs can be investigated to see whether what was displayed or was it recorded what was actually displayed on the screen of the individual terminals, as opposed to the centralised database? So, that is my first supplementary question.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates My second supplementary question is essentially on the nature of the question itself. It would seem that given that the Ministry of Education (MOE) is convinced that no actual variation in the question displayed happened, but about 8% to 9% of students who took the exam experienced this, would MOE consider varying the way they set oral examination questions in the future?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates I understand that reading comprehension is one of the elements being tested in an oral examination, but I would argue it is probably not the main element being tested. Would MOE consider in the future of setting questions or user testing questions such that such incidents would not arise again in the future and cause potential distress to our students?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, I thank Mr Low for the questions. On his first point, indeed, the investigation involved the individual terminals, including logging on with the student's ID and authentication, so that the exact same handshake between the terminal and the centralised database was carried out, to see what was returned as material. So, very thorough from a from a technical perspective.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates I would point out that his 8% to 9% is on that that day – I presume he is taking the 432 divided by 5,000 – but that was for that day only. The total number, of course, of people taking the examination across the four days is approximately four times as many.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Yes, indeed, SEAB will look to learn from this incident how we can reduce the risk of prompts being misread, guide the students about the preparation better, both in terms of the preparation for the exam and during the exam itself. And part of that technical professional skillset is to look at how specific phrases or words are used within examination prompts. SEAB will certainly do this.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Assoc Prof James Lim.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, I believe the incident reports of the anomaly were made by a number of individuals, not only by – and I have to be careful about the use of this word here – at least 432 students, but also by the teachers themselves.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates I accept that SEAB did their due diligence in verifying that there were no computer issues that could have given rise to the alleged outcome, and it is on this basis that the Senior Minister of State has asserted that that there were no errors. I wonder then is SEAB implicitly suggesting that there could have been an occasion of collective hallucination that occurred on that day? And if so, does this mean that the investigation would have perhaps implicitly privileged computer forensic evidence over the contemporary testimony of a significant number of human witnesses?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, implicit in Mr Lim's question, in what he describes as a collective hallucination, is that all the students recalled the same phrase. And as our investigations have demonstrated, the students demonstrated significant variance in the many phrases that they recalled. So, I think his implication that we are privileging one set of information over the other is incorrect.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates We are looking at both the testimonies, as well as the technical assessment. And the fact that the testimonies covered a wide variation in the phrases in itself is a very significant data point.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates So, I would ask him to perhaps not put words in my mouth about collective hallucinations, nor imply a privileging of one set of data over the other. I think in my response, I have explained how we take all the data very seriously.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Mr Kenneth Tiong.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Thank you, Speaker. Three supplementary questions.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates First, a factual question. Reports were filed on that day by SEAB's own examination personnel at the schools. So, how many irregularity reports did SEAB receive concerning 15 July English oral exam? Not complaints from candidates, but actual reports filed by its own oral examiners and presiding officers. And if the Ministry does not have this figure today, will they undertake to provide it? It seems very peculiar to me that 432 candidates from 132 schools reported it on day three, and day three alone. They did not report it on day one, day two or day four. So, to me it is highly anomalous.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Supplementary question two: did SEAB examine the candidates' contemporaneous physical planning sheets against the examiner's script?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates And supplementary question three: since SEAB has all the evidence needed – the login logs, the examiners' marking comments, the contemporaneous planning sheets, as well as the audio from the session – will the Ministry allow someone outside SEAB to look at them to conduct an external investigation?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, I do not have the numbers for the officers reporting. We can certainly look it up.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates We have examined the contemporaneous planning sheets. It is hard to correlate those with specific reports because many of the contemporaneous planning sheets did not have candidates' names written on them; that is not a requirement. But the information from the contemporaneous planning sheets does, in some cases indicate that the student had misread the prompt. And in some cases, it indicates that the student had misread the prompt, started planning on that basis and then corrected themselves. And so on one sheet, we can see the student having written down what they thought originally what is the prompt, and then crossing that out and then writing the correct prompt and then planning from there.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Indeed, we have had GovTech come in to audit the technical platforms, the processes and so forth. And they have come to the same conclusion as SEAB.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Mr David Hoe.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates I thank the Senior Minister of State for his response. My question comes from conversations with parents who accept the fact that their children might have misinterpreted or misread the question under examination pressure. So, three supplementary questions.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates First, is the ability to accurately process a question under exam pressure, part of a skill to be accessed under oral examination?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Second, if so, how are students prepared for this, because, unlike written exams, which students experience across many subjects, they may have fewer opportunities to practise this skillset to respond under exam pressure.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Third, parents would like to know, when candidates reported the question has been asked differently from what they saw in the terminal, could we clarify what was the process thereafter?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Sir, I thank Mr Hoe for his questions.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates His first two questions, I think, are a little bit outside the remit of this Parliamentary Question, which is about the examination process; and more about the overall preparation for the subject, the curriculum and the approach taken to prepare students. But across all subjects, certainly, the way in which our students approach an examination is very much part of the preparation in the classroom, in the development of the student across the curriculum, across the syllabus and the skills and techniques that they learn in the classroom. And some of that will be about how to deal with errors, deal with the emotional state that they are in. So, I think it is quite beyond the remit of the Parliamentary Question today, but I thank him for highlighting this. It is indeed an important part of the work that is done by the teachers.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates He then asked about the process after reporting. Can I just ask what is it he was asking about? I did not quite understand what he meant.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Mr Hoe.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Thank you to the Senior Minister of State. Basically, parents are questioning, after a child report that the question that they saw on the terminal and the question asked was different, what happened after?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates With the examiner or what?
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates With the exam at that point.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates Thank you, Sir. Mr Hoe has clarified that he is talking about what is happening with the examiner rather than the investigation of the incident.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates The examiners are trained in a process of positive marking. They are not looking for absolute compliance with a template answer. They are looking in this exam, for the students to demonstrate their skill, their oral skill with their language – what vocabulary they are using, their sentence structure, how they use grammar and tense, how they present their ideas, how they organise their ideas and how they build upon ideas over the course of the presentation.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates And this is one of the reasons why you can see that the marks are comparable. Because the students were able to speak. Sometimes, they had prepared incorrectly, but nevertheless were able to speak well, either on the initial subject that they thought they were presenting on or when corrected on the prompt that they heard from the examiner. And so, within their presentation was enough material demonstrating their ability with the language.
Probe into 15 July GCE "O" Level English Oral Exam Incident, Lessons Learnt and Support for Affected Candidates This approach of positive marking then allows the examiner to draw out the skills that the student has and provide them an appropriate mark. So, that is the process. And it applies regardless of whether the students hesitate, perhaps, ask a question about the prompt. There is this approach of positive marking: looking for the ability of the student to demonstrate the skills that they have, not blind compliance with a template, nor a negative marking approach looking for mistakes.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI 4 Mr Yip Hon Weng asked the Coordinating Minister for Social Policies and Minister for Health regarding the Singapore Medical Foundation AI Model (SIMFONI) initiative (a) what are the rationale and cost of developing AI models for conditions that can be diagnosed by routine tests, instead of targeting rare diseases that are harder to diagnose; (b) how will clinicians access SIMFONI and whether private practitioners will be included; and (c) what safeguards protect patient identifiers from data leaks.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI 5 Mr Low Wu Yang Andre asked the Coordinating Minister for Social Policies and Minister for Health (a) before the deployment of Singapore Medical Foundation AI Model (SIMFONI), whether the Ministry will subject its models to privacy testing by assessors independent of the developers to determine if they can memorise or reproduce identifiable patient information; and (b) whether test standards, findings and response protocol for any disclosure will be published.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Mr Speaker, may I have your permission to answer Question Nos 4 and 5 together?
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Yes, please proceed.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI The Singapore Medical Foundation AI Model (SIMFONI) develops artificial intelligence (AI) tools to provide clinical decision support for managing chronic conditions, prioritising conditions with high disease burden in Singapore and feasibility of development. We are starting with a couple of specialty areas and settings in the public healthcare sector. If successful, we will expand its coverage to more specialities and also possibly, to the private practitioners.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI SIMFONI's models are built on existing models trained using international data. They are adapted using local patient data, they are de-identified, in a secure national platform with well-established processes, expertise and safeguards. This platform is called TRUST, which was established in 2022 precisely for research, development and innovation efforts.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI To further ensure the integrity of this process, an independent safety and evaluation unit will carry out rigorous testing of SIMFONI's models before actual deployment.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Mr Yip.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Thank you, Mr Speaker. I thank the Minister for his response. While clinicians will ultimately retain responsibility for medical decisions, international experience has shown that doctors may gradually defer to AI decisions or recommendations – something known as automation bias. Will the Ministry announce specific safeguards to detect and mitigate this risk and how we ensure that AI compliments rather than subtly replaces independent clinical judgement?
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI We do not really announce what we are going to do to safeguard against this. This is very much in the way we approach AI, how we deploy and how we implement them.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI I have given a few speeches on this matter. There is a lot of hype about AI, including in healthcare. And sometimes, you really wonder why are these proposals like that. Because those of us who work in healthcare, you know it is not simple, especially a public essential service.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI We have to take a very careful use-case approach. That means, what problems are we facing, can AI solve those problems? Find the right tools. Never treat AI as a hammer going around looking for nails or a solution looking for a problem. We have enough problems. Let us see what AI can do to solve some of these problems.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI And when you can find the right AI tool, make sure it is trained in the local context, taking into account security, de-identified data, patient data that is in Singapore.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI When the model is ready, when you deploy it, you have to deploy in accordance with clinical protocols. There is a certain way we do things. There are existing legacy IT systems. AI is not something you plug into the socket and suddenly, it solves your problems. It does not work that way. And then, how do you interface with existing legacy IT systems? How the processes are thought through, with the clinicians still making the judgement?
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI It is early days, but I think we are very conscious that there is a lot of hype, a lot of use or misuse of AI. I think all this hype at some point will subside to a more realistic level, where people are more judicious, thoughtful and even, sober about using AI. And we tell ourselves, "Why wait for that process? Why do we not start off being sober and judicious?" And that is the approach that the Ministry of Health is taking.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Mr Low.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Thank you, Speaker. Minister, I have three supplementary questions. Firstly, it is about the scope of the data that is being envisaged to be used to train this AI model. Would it encompass all health data in Singapore of all patients, including data in the National Electronic Health Record (NEHR) database?
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI Secondly, it is about the nature of the consent that will be sought before patient's data will be used for training. Has the Ministry considered, or will the Ministry consider, asking patients for explicit consent before their data can be used for training purposes?
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI And third supplementary question: I understand the Minister has given assurances that the data will be anonymised and have personal identifiers removed before they are used for training. However, as we can see, I have another Parliamentary Question today for a similar data leak incident with the Singapore Land Authority, where a lot of these processes are fallible. And sometimes, anonymisation of data does not actually pan out and personal identifiable data can leak into publicly available databases. What assurances can the Minister give to the public that their very sensitive health data will remain safe?
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI I think we all know AI is a breakthrough technology, but we must take a thoughtful and judicious approach. So, we have to decide what we want to do with it. If we start off thinking that we distrust this system entirely and therefore, let us scope down the data to be used to train models – even though we know the data is essential to train models that can improve healthcare, save lives, operate everything better and maybe help us address some of our very stark challenges of an ageing population with rising patient load – and if we accept that, but yet at the same time distrust the system and say, "Let's scope down the data, let's not believe that they can really be anonymised and de-dentified because there can be leaks and traceability", and "Let's seek everyone's consent before we can use", then it is a non-starter. It becomes a non-starter.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI I think just take a balanced approach: recognise that this is a breakthrough technology that can do a lot of good, but to do so, we will need to use the data that we have and take the necessary precautions. Make sure that it is anonymised and de-identified. And we started this process some years ago. So, TRUST data has been used in various contexts already – so far, incident-free. I think we make sure it is anonymised and de-identified by international standards.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI I think let us proceed on that basis. We must have safeguards, I agree. But we must strike a balance between leveraging that technology, but putting in place the safeguards. And we can proceed on that basis.
Rationale and Cost of Developing AI Models for Healthcare Diagnostics, and Safeguards for Patient Data - Update on deployment of SIMFONI And at some point, Singaporeans will see that healthcare services are improving, yet quite incident-free. That is when they have confidence and assurance. I think let the action and the results speak for themselves.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight 6 Ms Hazlina Abdul Halim asked the Acting Minister-in-charge of Muslim Affairs in light of recent disruptions affecting korban services in Singapore, whether MUIS will (i) consider reviewing the current operating model for korban services in Singapore for better oversight (ii) introduce a licensing or accreditation framework for such operators and (iii) establish a service standard for local operators to adhere to.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight Mr Speaker, a question on korban disruption during this year's Hari Raya Haji was earlier addressed at the last Parliamentary Sitting on 7 July 2026. Nevertheless, I felt that it was important to address this matter again today because korban is a significant religious observance for our Muslim community.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight To recap, on 21 May this year, the Islamic Religious Council of Singapore (Majlis Ugama Islam Singapura or MUIS) was informed of a delay in the arrival of 400 livestock, which affected local korban services at four local mosques. This occurred because a local korban vendor was unable to secure the necessary export approvals from the Australian authorities in time.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight The incident had caused concern and disappointment, understandably, to those who had made korban arrangements at these mosques. At the same time, I was heartened by the way the community responded. Many showed patience, understanding and support during this period, and this reflects the spirit of the community in supporting one another through difficult situations.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight MUIS has conducted a comprehensive review of the incident. In the two weeks since I have stepped into this role, I have discussed the findings and action plans with MUIS, and we will implement concrete measures to strengthen the current operating model and better support local korban operators.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight Let me briefly recap three key measures that MUIS will implement.
Reviewing Operating Model for Korban Services in Singapore for Better Oversight First, MUIS will strengthen oversight of local korban operators, particularly for new vendors. We will work closely with vendors to ensure that necessary overseas regulatory approvals are obtained in a timely manner and that crucial milestones for korban operations are met.