In short
On 13 February 2026, the Well being Sciences Authority (HSA) printed the findings of its public session on the proposed exemption from producer’s licensing and product registration necessities for synthetic intelligence software program as a medical system (AI-SaMD) developed by chosen public healthcare entities to be used in public healthcare (AI‑SaMD sandbox).
The general public session was held from 19 Might 2025 to 19 June 2025, and the HSA introduced that the suggestions was usually supportive of the proposed AI-SaMD sandbox. The HSA additionally printed a abstract of the strategies offered.
In additional element
As coated in our Might 2025 alert, the HSA invited the general public to offer suggestions on the proposed AI-SaMD sandbox from 19 Might 2025 to 19 June 2025.
The respondents usually expressed assist for the proposal and offered constructive strategies, which broadly fell into the next classes:
- Strategies to strengthen controls throughout the sandbox, similar to imposing extra Worldwide Organisation for Standardisation (ISO), Worldwide Electrotechnical Fee and cybersecurity necessities, and for the AI-SaMDs within the sandbox to finally be registered with HSA
- Strategies to broaden the scope of the AI-SaMD sandbox to permit different healthcare professionals, other than registered medical practitioners, to offer oversight over the event of those AI-SaMDs, and lengthening participation past the general public healthcare sector, similar to to non-public healthcare establishments and software program builders
The HSA emphasised that the proposed sandbox is meant to make sure applicable regulatory oversight, whereas on the identical time fostering healthcare innovation. Subsequently, the HSA has to think about the prevailing cybersecurity and information safety safeguards that public healthcare entities must adjust to, in addition to the prevailing necessities underneath the Well being Merchandise Act and the Healthcare Providers Act.
Bearing in mind the above issues, the HSA will undertake the next:
- Scope the sandbox solely to AI-SaMDs that pose decrease potential affected person dangers, i.e., Class A and Class B AI‑SaMDs, which solely purpose to diagnose and/or drive scientific affected person administration for non-critical medical situations (For extra info on the danger classification of AI-SaMDs, see our earlier alert.)
- Require a clinician employed in a public healthcare establishment, and holding the place of guide or larger, to supervise the AI-SaMD’s design, validation and output, and to make sure that the AI-SaMD is manufactured underneath the supervision of a certified medical practitioner
- Require public healthcare builders to conduct yearly high quality administration system self‑attestations to substantiate compliance with ISO 13485 requirements
- Require AI-SaMDs to be endorsed by the chair of the medical board or the CEO of the general public healthcare establishment for deployment, guaranteeing that senior management is conscious of and accountable for the AI-SaMDs being deployed of their public healthcare establishment or cluster
- Require that the general public healthcare establishment or cluster notify the HSA of the AI-SaMDs earlier than deploying them to be used in sufferers
- Require that deployers inform sufferers when AI-SaMDs developed underneath the proposed sandbox will likely be used of their medical care
- Interact with public healthcare builders on the HSA product registration upon the broader deployment of the AI-SaMD throughout public healthcare
The HSA reiterated that the sandbox is an enlargement of the exceptions throughout the Well being Merchandise Act, and the HSA will proceed to uphold the important thing safeguards, similar to post-market obligations and accountability, to make sure regulatory oversight and affected person security.
Key takeaways
The HSA has been embracing and inspiring the adoption and use of AI throughout the Singapore healthcare system, together with offering regulatory steerage to make sure affected person security. This AI-SaMD sandbox is one such measure that facilitates broader AI adoption and scaling up throughout the Singapore healthcare ecosystem, and the HSA has confirmed that it will likely be contemplating extending the scope to past public healthcare.
The HSA will intently monitor the outcomes of the sandbox to guage and decide the following steps, together with the calibration of controls, and to create a complete framework to assist AI innovation whereas guaranteeing the secure and efficient use of AI in healthcare.
Firms ought to proceed to observe how the HSA calibrates its controls towards AI-SaMDs in Singapore, because the outcomes of the sandbox might have knock-on results on wider controls and steerage for AI-SaMDs.
In short
On 13 February 2026, the Well being Sciences Authority (HSA) printed the findings of its public session on the proposed exemption from producer’s licensing and product registration necessities for synthetic intelligence software program as a medical system (AI-SaMD) developed by chosen public healthcare entities to be used in public healthcare (AI‑SaMD sandbox).
The general public session was held from 19 Might 2025 to 19 June 2025, and the HSA introduced that the suggestions was usually supportive of the proposed AI-SaMD sandbox. The HSA additionally printed a abstract of the strategies offered.
In additional element
As coated in our Might 2025 alert, the HSA invited the general public to offer suggestions on the proposed AI-SaMD sandbox from 19 Might 2025 to 19 June 2025.
The respondents usually expressed assist for the proposal and offered constructive strategies, which broadly fell into the next classes:
- Strategies to strengthen controls throughout the sandbox, similar to imposing extra Worldwide Organisation for Standardisation (ISO), Worldwide Electrotechnical Fee and cybersecurity necessities, and for the AI-SaMDs within the sandbox to finally be registered with HSA
- Strategies to broaden the scope of the AI-SaMD sandbox to permit different healthcare professionals, other than registered medical practitioners, to offer oversight over the event of those AI-SaMDs, and lengthening participation past the general public healthcare sector, similar to to non-public healthcare establishments and software program builders
The HSA emphasised that the proposed sandbox is meant to make sure applicable regulatory oversight, whereas on the identical time fostering healthcare innovation. Subsequently, the HSA has to think about the prevailing cybersecurity and information safety safeguards that public healthcare entities must adjust to, in addition to the prevailing necessities underneath the Well being Merchandise Act and the Healthcare Providers Act.
Bearing in mind the above issues, the HSA will undertake the next:
- Scope the sandbox solely to AI-SaMDs that pose decrease potential affected person dangers, i.e., Class A and Class B AI‑SaMDs, which solely purpose to diagnose and/or drive scientific affected person administration for non-critical medical situations (For extra info on the danger classification of AI-SaMDs, see our earlier alert.)
- Require a clinician employed in a public healthcare establishment, and holding the place of guide or larger, to supervise the AI-SaMD’s design, validation and output, and to make sure that the AI-SaMD is manufactured underneath the supervision of a certified medical practitioner
- Require public healthcare builders to conduct yearly high quality administration system self‑attestations to substantiate compliance with ISO 13485 requirements
- Require AI-SaMDs to be endorsed by the chair of the medical board or the CEO of the general public healthcare establishment for deployment, guaranteeing that senior management is conscious of and accountable for the AI-SaMDs being deployed of their public healthcare establishment or cluster
- Require that the general public healthcare establishment or cluster notify the HSA of the AI-SaMDs earlier than deploying them to be used in sufferers
- Require that deployers inform sufferers when AI-SaMDs developed underneath the proposed sandbox will likely be used of their medical care
- Interact with public healthcare builders on the HSA product registration upon the broader deployment of the AI-SaMD throughout public healthcare
The HSA reiterated that the sandbox is an enlargement of the exceptions throughout the Well being Merchandise Act, and the HSA will proceed to uphold the important thing safeguards, similar to post-market obligations and accountability, to make sure regulatory oversight and affected person security.
Key takeaways
The HSA has been embracing and inspiring the adoption and use of AI throughout the Singapore healthcare system, together with offering regulatory steerage to make sure affected person security. This AI-SaMD sandbox is one such measure that facilitates broader AI adoption and scaling up throughout the Singapore healthcare ecosystem, and the HSA has confirmed that it will likely be contemplating extending the scope to past public healthcare.
The HSA will intently monitor the outcomes of the sandbox to guage and decide the following steps, together with the calibration of controls, and to create a complete framework to assist AI innovation whereas guaranteeing the secure and efficient use of AI in healthcare.
Firms ought to proceed to observe how the HSA calibrates its controls towards AI-SaMDs in Singapore, because the outcomes of the sandbox might have knock-on results on wider controls and steerage for AI-SaMDs.
