[Introductory sections omitted. Footnotes, figures and anecdotal/case study insets omitted throughout; refer to official version.]

NATIONAL AI STRATEGY 2.0

Our Vision and Goals

Developed through extensive engagements, NAIS 2.0 starts with the conviction that we must do our utmost to harness AI for the Public Good, for Singapore and the World. Singapore will be a place where AI can:

-	Address the needs and challenges of our time. For example, in areas of global importance such as population health and climate change.

-	Be the great equaliser. We shall uplift and empower our people and businesses, equipping them with the capabilities and resources to thrive in an AI-enabled future.

NAIS 2.0 seeks to attain the twin goals of:

Excellence. We will selectively develop peaks of excellence in AI, to advance the field and maximise value creation.

Empowerment. We will raise up individuals, businesses, and communities to use AI with confidence, discernment, and trust.

Our Plans

To achieve our vision and goals, we will direct efforts under NAIS 2.0 toward three Systems, working through 10 Enablers.

-	System 1: Activity Drivers (Enablers: Industry, Government, Research). Industry, Government, and public research performers have deep technical capabilities that can be applied to deliver value. We need to orchestrate them around meaningful use cases and problem statements to transform our economy and society.

-	System 2: People & Communities (Enablers: Talent, Capabilities, Placemaking). We will attract more top-tier researchers and engineers to work with and from Singapore. More of our technology workforce should work to scale novel AI solutions, that form part of the toolkit which a confident base of enterprises and workers can use.

-	System 3: Infrastructure & Environment (Enablers: Compute, Data, Trusted Environment, Leader in Thought and Action). We will ensure that Singapore hosts the necessary infrastructure and provides a trusted environment for AI innovation. This will make us a credible leader and preferred site for AI development, deployment, and adoption.

This strategy statement outlines 15 Actions that Singapore will undertake across these systems and enablers, to support our ambitions over the next 3-5 years. We will continually review these Actions to respond to fast-moving developments in AI, across domains.

 

SYSTEM 1: ACTIVITY DRIVERS

Industry, Government, and Research drive AI activity, from basic science to product R&D and adoption. Together, they help Singapore leverage the rapid innovation cycles in AI for the economy and our society.

Singapore has built up a strong and promising base of AI capabilities, which we should harness to significantly uplift the broader ecosystem. We will encourage greater experimentation and collaboration, focusing on interesting and impactful challenges, to reap greater benefits from AI.

We shall steer AI toward developing select peaks of excellence, which can deliver outsized impact to Singapore and the lives of Singaporeans.

-	AI in Domains. We will encourage AI innovation and adoption in key domains, including:

*	Leading Economic Sectors, which form a sizeable share of Singapore’s real Gross Domestic Product (GDP) and for which AI innovation could catalyse the next bound of economic growth. Examples include Manufacturing, Financial Services, Transport & Logistics, and Biomedical Sciences.

*	Smart Nation Priorities, where AI assists our national development, and unlocks new value propositions for social impact. These include Healthcare, Education & Manpower, Trust & Safety, and Public Service Delivery.

-	Cross-cutting Capabilities. We will invest in cross-cutting areas that can accelerate the development and deployment of AI solutions across all domains, including:

*	AI for Business Operations, where AI can optimise and transform business functions (e.g. customer relationship management, finance, human resources, legal, sales and marketing, and supply chain management).

*	AI for Science, where the use of AI can accelerate research productivity across scientific domains (e.g. drug discovery).

*	Foundational AI, where scientific advancements are still needed to improve AI’s abilities. For a start, we will double down on three specific themes – Reasoning AI, Resource-Efficient AI, and Responsible AI³ – with a view to making AI less costly, more widely-used, and most importantly, trusted.

 

ENABLER 1: Industry

Singapore’s knowledge-intensive economy is especially exposed to AI. Our industries and enterprises need to be responsive, and ready to create AI-driven opportunities that will strengthen Singapore’s economic competitiveness. This will help workers and industries set the pace and shape their own future, rather than face untimely disruption.

Many companies have already embarked on AI-enabled innovation projects, which are often focused on the Application layer. While these are useful and have made a difference, there is scope for us to go further. Deep and transformative AI innovation today increasingly requires integrated capabilities from across the entire technology stack (i.e. including Model and Infrastructure layers). There must also be a stronger nexus between sophisticated end-users and leading-edge AI innovators and producers.

To anchor transformative AI innovation and value creation in Singapore, the Government has a key role in curating the right incentives (e.g. targeted grants), resources (e.g. compute and talent), regulatory frameworks, and partners. For a start, we will adopt a sector-specific, use case-centric approach, and we will focus our initial efforts on Leading Economic Sectors assessed to be most ready for AI-driven transformation.

Done well, we will have a thriving industry ecosystem in Singapore, with significant value creation from AI, and capabilities across the AI technology stack. AI will complement our workforce, boost our productivity, and be a differentiating factor in attracting best-in-class companies to Singapore. We will also seed a virtuous cycle where industry end-users can tap into a dense network of AI producers, who in turn are motivated by the presence of strong, industry-relevant lead demand and product mandates.

Action 1

Anchor new AI Centres of Excellence (CoEs) across companies, and explore establishing Sectoral AI CoEs to drive sophisticated AI value creation and usage in key sectors

We will attract and anchor new AI CoEs in Singapore-based companies that are leading-edge producers (e.g. technology companies, start-ups) and sophisticated end-users, in order to conduct value creation activities across the AI stack. These AI CoEs will (a) concentrate and deepen companies’ innovation capabilities across the AI stack; (b) own core AI product charters and functions; and (c) align with the peaks of excellence that Singapore aims to build. Their roles could include creating new intellectual property, products, and services, beyond optimising business processes.

In addition, we will explore establishing a new model of Sectoral AI CoEs to intensify sophisticated AI value creation and usage in selected economic sectors. We will work with Industry Champions⁴ to identify sector use cases and crowd in a broader base of researchers and companies in the sector.

⁴ Industry Champions refer to enterprise leaders that can form lead demand for AI solutions, boost their sector’s competitiveness, uplift the capabilities of their supply chains, and generate spin-offs for the broader sector ecosystem.

Action 2

Strengthen our AI start-up ecosystem, including attracting AI-focused accelerator programmes to spur rapid AI experimentation

In concert with private sector partners, we will strengthen the AI start-up ecosystem along several pillars. This includes attracting more venture builders and developing more accelerator programmes. We want to speed up AI value discovery across industry, and nurture a pipeline of disruptive, AI-native start-ups. These accelerator programmes could be led by a mix of big technology companies and venture capital firms. They will provide the capital, business and technical expertise, infrastructure, and market networks to spur rapid AI experimentation. We will nurture globally-oriented AI innovators to create intellectual property and scale to more markets.

 

ENABLER 2: Government

The Government is committed to the continuous improvement of public service delivery for our people and businesses. We will harness AI to serve the public in more impactful ways.

We also recognise the positive spin-offs from such efforts. These include providing lead demand for commercial AI tools, demonstrating AI-led transformation for the rest of the economy and society, spurring private sector investment in AI, and mainstreaming AI adoption.

With this, citizens can easily provide municipal feedback without having to figure out which agency they should contact. Agencies can also attend to cases more quickly.

For the Government to better harness AI, we need to:

-	Develop, deploy, and integrate more useful and powerful AI-enabled products. These include customised or commercial off-the-shelf solutions. They can support the Government’s needs, both general (e.g. smart transcription tools for citizen support services) and specialised (e.g. detecting anomalies in financial statements and transactions to assist anti-money laundering efforts).

-	Drive mass awareness and adoption of AI across the Public Service. This will encourage public officers to use AI-enabled products and services confidently, for more efficient and effective public service delivery.

-	Work with industry and the public to better identify use cases and address pain points using AI. This can be done through well-structured ideathons, hackathons, and incubator programmes.

Action 3

Improve Public Service productivity, with new value propositions for our citizens

The Government will accelerate public sector adoption of AI to unlock new value propositions for our people and businesses. This will take place at two levels:

-	Smart Nation Priorities. Government agencies equipped with the specialised knowledge, technical capabilities, and regulatory tools for these domains (e.g. Healthcare and Education) will develop and lead sector-specific AI strategies, to address the needs and challenges of these domains.

-	Whole-of-Government functional domains. These include Finance, Human Resources, and Service Delivery, which are key to transforming government processes and services for the better. All Government agencies that lead such functions will identify and optimise specific business lines with AI, and build internal capabilities to deliver business outcomes.

The Government will coordinate the resources needed to support public sector AI adoption. For example, through central funding to support or scale-up novel and impactful use cases, facilitating cross-agency data sharing, providing access to high-performance compute and associated engineering capabilities (e.g. through AI Trailblazers), and issuing facilitative policy guidelines.

We will also uplift the AI literacy baseline and sharpen the AI proficiency of all public officers. We will develop and roll out targeted courses to impart AI skills, for different public sector workforce segments (e.g. senior management, policy, and operational roles).

 

ENABLER 3: Research

A strong community of scientists and researchers contributes to a vibrant AI ecosystem in Singapore.

-	They add to our overall technical heft, making us attractive to other top researchers, investors, and entrepreneurs, to create new value in AI. This can generate other economic spin-offs, such as the creation of well-paying jobs, start-up growth, and sophisticated venture capital investments.

-	Their research breakthroughs can also translate into commercial products, which can be used by our companies and across government agencies to realise better outcomes.

-	Their presence helps secure an enduring competitive advantage in AI for Singapore. Over the last few decades, innovators in Silicon Valley and other AI research hubs (e.g. Beijing, Boston, Montreal, Seattle, and Toronto) have led these places’ technological prowess. Singapore will similarly need strong research credentials.

Singapore will take a pragmatic yet bold approach to building up and sustaining our leadership in select areas of AI research.

Action 4

Update national AI R&D plans to sustain leadership in select research areas

We intend to update our national AI R&D plans in five ways: Research Priorities, Industry-Academia Nexus, Talent, Compute, and International Collaborations.

-	Research Priorities. To optimise our limited resources for impact, we will be selective in our AI research priorities, and align them to the peaks of excellence that Singapore hopes to build.

-	Industry-Academia Nexus. We will foster more R&D collaboration between academia and industry, capitalising on the growing porosity of talent and innovation activities across the two spheres.

-	Talent. We will strengthen efforts to recruit top AI researchers to work from and with Singapore (see Action 5).

-	Compute. We will secure access, and operate Graphics Processing Units (GPUs) for Singapore’s research community. This will also build up Singapore’s infrastructure engineering expertise.

-	International Collaboration. We will expand international research collaboration in areas aligned with our research priorities. This could take the form of joint grant calls or PhD training programmes with other countries, or involvement in international conferences. This will allow Singapore to level up our current capabilities by working with the best in the world and make a meaningful contribution to global AI development.

 

SYSTEM 2: PEOPLE & COMMUNITIES

Tight-knit knowledge communities are critical for AI innovation, driving the exchange of ideas and expertise for research, product development, and the impactful use of AI. Singapore must be home to exemplary AI Creators, Practitioners, and Users. We must:

-	Attract, anchor, and develop more AI Creators (top-tier AI talent) from industry and academia to work with and from Singapore, to generate cutting-edge AI research and drive novel use cases.

-	Increase the number of AI Practitioners (tech workers) with the skillsets to create, implement and deploy AI systems, models, and algorithms in organisations, at scale.

-	Build up a base of confident AI Users (enterprises and general workforce) that are equipped to use AI-powered products and services to increase productivity, and pursue better jobs and more impactful work.

-	Accelerate the exchange of ideas, both within our local AI community of Creators, Practitioners, and Users, and across global networks.

 

ENABLER 4: Talent

Singapore will expand the pool and raise the quality of our AI Creators and Practitioners, to accelerate AI innovation and support AI Activity Drivers.

-	We must increase the number and diversity of AI Creators working from and with Singapore, to drive leading-edge R&D activities and product development.

-	We must also boost our pipeline of AI Practitioners. They are critical to supporting top-tier AI activity, translating innovation into products and services, and transforming our industries.

Action 5

Attract world’s top AI Creators to work from and with Singapore

Singapore will engage and attract world-class AI Creators, from both the public research and industry spheres, to deepen their innovation activities here. We will:

-	Set up a dedicated team to identify, engage, and anchor AI Creators. This team will be the primary interface to facilitate the integration of AI Creators into the Singapore ecosystem, including through bespoke support mechanisms.

-	Create novel value propositions to attract AI Creators. We will intensify local AI development activities and induct international experts into our AI ecosystem to create the building blocks for success. We will explore possible modalities for doing so, such as hybrid Singapore-overseas working arrangements, part-time appointments across industry and academia, and partnerships with international research institutions or companies.

Action 6

Boost AI Practitioner pool to 15,000

Singapore will boost the pool of AI Practitioners to support growing AI demand, by:

-	Scaling up AI-specific training programmes. We will re-design the AI Apprenticeship Programme (AIAP)⁵ to significantly increase the number of apprentices we can train annually. We will also work with industry AI product development teams to expand the number of company attachments for our Continuing Education and Training programmes.

-	Scaling up technology and AI talent pipelines, through Pre-Employment Training and by reskilling and upskilling workers through Continuing Education and Training.

-	Remaining open to global tech talent. We will continue to welcome global AI talent to work and live in Singapore, and contribute to our national efforts.

 

ENABLER 5: Capabilities

It is important to uplift the AI capabilities of our industries and workforce, to enable our enterprises and workers to reap the benefits of AI, and minimise its disruptive effects.

-	AI has the potential to help enterprises increase their productivity and stay relevant, provided they have the capabilities to apply it well.

-	Workers must also be equipped with the necessary skillsets to utilise AI tools, so that they can increase their productivity, avoid untimely disruptions to their jobs, and help enterprises optimise their operations and improve their competitiveness.

While all economies must prepare for AI disruptions, we believe that Singapore is well-placed to navigate them. Singaporeans have persistently demonstrated resilience and adaptability in the face of change, and the Government has a strong track record of managing such nation-wide transitions, whether structural or cyclical, through our policies and outreach. We will build upon our existing toolkits, schemes, and training programmes, to empower enterprises and workers to adopt AI and stay ahead of the curve.

Action 7

Intensify enterprise AI adoption for industry transformation

Singapore will intensify the promotion of AI adoption across all enterprises.

-	We have made available tools that enterprises can use to evaluate their readiness to adopt AI. For example, self-assessment tools like AISG’s AI Readiness Index (AIRI) are useful in helping companies identify and map out the capabilities and infrastructure they need for using AI.

-	We will promote baseline digital adoption for our enterprises and enhance the enterprise digitalisation toolkit, to support more sophisticated AI adoption. Businesses’ first interactions with AI often happen through their existing enterprise solutions. Only after they have had the foundational capabilities in place, and are convinced of AI’s benefits, do they then turn to more advanced or bespoke AI business solutions. Through programmes like SMEs Go Digital and CTO-as-a-Service (CTOaaS), the Government has helped businesses gain awareness of suitable AI-enabled solutions. We have also encouraged the adoption of sector-relevant AI solutions, through our refreshed Industry Digital Plans (IDPs).

-	For more digitally mature enterprises, we will provide tailored support for AI-enabled business transformation. This includes leveraging IMDA and EnterpriseSG’s Digital Leaders Programme (DLP), which helps companies build in-house digital capabilities, as well as the scoping of impactful AI projects in partnership with AI solution providers.

Action 8

Upskill workforce through sector-specific AI training programmes

To help workers acquire the necessary skillsets to work with AI, Industry Transformation Map (ITM) Sector Leads will develop targeted interventions for their sectoral workforce. AI has multiple applications in different contexts, and ITM Sector Leads are best placed to size the potential impact of AI on their respective industries, and by extension their workforce. Sector Leads can ride on existing frameworks like the Jobs Transformation Maps (JTMs) to identify the relevant skillsets needed, and develop appropriate training programmes for AI upskilling and reskilling.

 

ENABLER 6: Placemaking

In a vibrant ecosystem, diverse talent can readily connect with one another and forge meaningful partnerships. There are many talented AI Creators and Practitioners today, with good ideas that can expand the frontier of AI. The opportunity to spar and collaborate with like-minded peers can enrich these ideas, and accelerate the translation into products and new value. Such synergies are seen in global AI hubs such as San Francisco, where stakeholders working across all parts of the AI ecosystem are found in close proximity, and the vibrancy of the community in turn attracts the participation of even more talented individuals, companies, and capital.

To realise similar benefits, Singapore intends to provide more platforms which can bring our AI community together. We want to engage with more of our talent pool, and connect them to global AI experts for greater opportunities to interact and collaborate. Over time, we hope these connections will create a sense of identity and fraternity, and build up a broader Singapore AI community that can attest to the support and inspiration that this network provides.

Action 9

Establish an iconic AI site to co-locate AI creators and practitioners, and nurture the AI community in Singapore

Singapore will establish a dedicated physical place for AI in Singapore. This will be a focal point for our community of AI Creators and Practitioners to form new connections and spark new ideas:

-	This place will serve as an intellectual home for both Singapore-based individuals and visiting colleagues. We will welcome entrepreneurs, researchers, engineers, apprentices, and students, who are passionate about AI and share our mission orientation, to this place. There, we will include a variety of purpose-built spaces aimed at building up a sense of community and accelerating the exchange of ideas. We will co-create this place in close consultation with AI ecosystem representatives.

 

SYSTEM 3: INFRASTRUCTURE & ENVIRONMENT

Successful AI value creation requires robust and conducive infrastructure and a facilitative environment, where all stages of the AI life cycle are well supported, and everything works the way it should.

For Singapore, this will involve:

-	Availing compute and data for AI innovation. These are important building blocks to serve our AI ambitions.

-	A trusted environment where AI-enabled innovations and systems are robust and safe, so that our people can engage with AI with confidence.

-	Safeguarding Singapore’s AI interests in the international arena. We hope to participate in shaping the international rules of the road that are emerging around AI.

 

ENABLER 7: Compute

The increasing scale and proliferation of AI models have driven exponential growth in demand for chips that can support AI workloads (compute), such as GPUs. However, GPUs are in short supply, and we face intense global competition to access them. In addition, the resource-intensive nature of AI workloads requires adequate and sustainable infrastructure capacity, which we have to prepare for.

To support high value AI activities, Singapore must ensure reliable, localised access to high-performance compute, so that industry, academia, and the Government have the means to innovate and build here.

-	Careful management is needed, as this infrastructure will take up significant power, carbon, water, and land, all of which are limited in Singapore.

-	While most of the compute supply is expected to go toward industry use, we will direct a small proportion towards meritorious use cases that build up our local research and industry capabilities, or are in service of the Public Good.

We will also work with sustainability leaders in this space, to help Singapore achieve our AI ambitions while meeting our sustainability goals.

Action 10

Significantly increase high-performance compute available in Singapore

We will actively crowd in a significant amount of compute to Singapore to support our growing AI needs, especially in areas of national interest. To this end, we will:

-	Deepen our substantive partnerships with major compute players, ranging from chipmakers to Cloud Service Providers (CSPs), to secure local access to compute capacity.

-	Support Singapore-based compute with the required resource envelope. We will ensure that there is sufficient carbon budget and power allocated to support data centres, that house GPUs or their equivalents, to process AI workloads in the near term. In the medium to longer term, we will chart a roadmap towards the growth of net-zero, green data centres that are powered by renewable energy. This includes continued collaborations with industry on innovative proposals to push the sustainability envelope.

-	Manage a small subset of Singapore-based GPUs or their equivalent, to support meritorious use cases for capability building, innovation, and the Public Good.

 

ENABLER 8: Data

Data remains one of the essential factors of production for AI. Today, there are more avenues for AI development to overcome the traditional obstacles of data availability and access. The open-source movement has made large quantities of data available, while private sector data marketplaces also facilitate the commercial trading of datasets. Meanwhile, it has become easier to create and augment structured datasets for AI, including through the use of synthetic data.

National efforts to maximise value from data should therefore pivot away from access and availability, to address other emerging factors. These include improving the quality of datasets, and ensuring that data use for AI development is context-appropriate and operates within trusted data sharing frameworks. We are also keen to make more government datasets available to solve AI problems of high national priority. Done well, we will demonstrate to AI developers that Singapore is a conducive place for responsible AI development.

-	We want to uncover and avail more good data that AI Activity Drivers can leverage. While Singapore may not have the biggest datasets where size of population is the key determinant, the breadth and depth of our economic base suggest that meaningful datasets can still be mined to generate useful insights and derive new value.

-	Singapore should continue to lean forward to support industry efforts to employ data for AI. Importantly, given increased concerns around data security, we should invest in Privacy-Enhancing Technologies (PETs), and other novel approaches that can address barriers around data protection and sharing, to encourage more data flows.

-	The Government should also support the data needs of meritorious use cases that serve the Public Good, particularly if these are aligned with our national priorities but are not commercially attractive. This will ensure that such projects do not fall through the cracks.

Action 11

Build capabilities in data services and Privacy-Enhancing Technologies

Singapore will build up our data management capabilities so that more of us can store, manage, share, and govern the data that we have, enabling us to draw new insights and uncover more value. To do so, we will need to harness new technologies that will enable safe and trusted data sharing, and grow more capabilities to do so at scale.

-	We will develop capabilities in PETs. PETs enable businesses to access other datasets in a privacy-preserving manner, hence extending the pool of data from which they can derive insights. The Government will support research and development for PETs, especially in targeted areas like synthetic data generation, data annotation, federated learning, and homomorphic encryption, and expand practical measures like regulatory sandboxes and guidelines to promote experimentation with PETs.

Action 12

Unlock Government data for use cases that serve the Public Good

Singapore will selectively unlock more public sector data for AI development that serves the Public Good. We will assess the feasibility of setting up a “data concierge” within the Government, to engage AI Activity Drivers and Creators, and identify and facilitate access to the datasets that they require for their AI development activities.

-	For public sector datasets, the data concierge will help AI Activity Drivers and Creators discover and obtain access to appropriate public sector datasets. Where these are not already open-access, the data concierge will help unlock closed datasets through brokering data sharing agreements. It could also work upstream with agencies to collect data based on the client’s needs. This will be underpinned by a data-sharing evaluation framework that takes the “Public Good” dimension into account.

-	For private sector datasets, the Government can consider stepping in for selected high-value use cases where we can play a catalytic role. In such cases, the data concierge can explore facilitating handshakes between private industry entities as a trusted intermediary, to aggregate and produce data derivatives.

We will continue our thought leadership around the progressive use of data for AI development, such as advocating for trusted cross-border data flows and providing guidelines to businesses to educate and facilitate greater data use.

 

ENABLER 9: Trusted Environment

Trust and safety underwrite confidence in Singapore’s AI landscape. They lie at the heart of our interventions to deliver AI for the Public Good.

-	We will endeavour for AI to be developed and deployed in a safe, trustworthy, and responsible manner. The Government will institutionalise appropriate governance and security frameworks for AI systems. The ultimate aim is to establish a trusted environment for AI, where people can have the confidence that their interests are protected when interacting with AI.

-	We must retain agility in our regulatory approaches. AI will continue to evolve, and no party has full sight of the risks that might emerge. It is only through experimentation and exploration that the AI community can deepen its understanding of AI, and discover and address potential risks. The Government must therefore take a pragmatic approach – supporting experimentation and innovation, while still ensuring that AI is developed and used responsibly, in line with the rule of law and the safeguards we have put in place. Where existing regulatory frameworks need to be updated, we will do so thoughtfully and in concert with others, accounting for the global nature of AI.

There is a range of potential risks around AI, spanning from concerns around model quality and fair use, to fears around the loss of control and existential risks. Singapore remains open to engaging with all perspectives, to enhance our understanding of the risk landscape and to inform our most urgent priorities for risk mitigation. For a start:

-	We must ensure that AI systems are well-developed, reliable, and resilient. This requires paying close attention to the model development process, to ensure that the output of models is not biased, inaccurate, or erroneous. AI models should also be aligned with the appropriate set of human and cultural values. Existing international conversations around AI governance and safety are largely centred on these concerns.

-	We must prevent AI models from being used in malicious or harmful ways, and secure them against adversarial attacks. If AI is used carelessly, it can potentially amplify negative outcomes like discrimination, anti-competitive behaviours, or intellectual property infringement. It can also be deliberately misused against us, to supercharge existing threats (e.g. scams, cyber-attacks, and mis/disinformation) in terms of scale, speed, and sophistication. We must remain vigilant against these risks to maintain digital trust.

The Government will take differentiated approaches to managing risks to and from AI, ranging from regulatory moves to voluntary guidelines, recognising that AI will continue to evolve. We will need a deeper understanding of how AI works, what benchmarks to use, and what testing is appropriate, and we look forward to developing these perspectives together with other stakeholders.

Action 13

Ensure fit-for-purpose regulatory environment for AI

The Government will continue to maintain a regulatory environment for AI that is pro-innovation while ensuring appropriate guardrails. We will:

-	Update AI governance frameworks to address novel risks. We will regularly review and adjust frameworks like the Model AI Governance Framework and AI Verify to reflect emerging principles, concerns, and technological developments (e.g. Generative AI). As part of this, it will be important to establish clear responsibilities for actors across the AI supply chain. This baseline guidance will give clarity to AI developers and users on how to be responsible in the design and use of AI.

-	Continue working with partners on R&D, particularly around alignment and evaluations. We will work with partners to translate guidelines into appropriate technical standards, tools, and services that can be practically applied. These will also be supported by policy measures including regulatory sandboxes, pilots for solutions such as watermarking and model cards, and capability development to nurture a domestic Testing, Inspection, and Certification (TIC) sector.

-	Design interventions that are risk-based, tiered, and adapted for specific vertical sectors and horizontal applications. This recognises that every use case carries a different set of considerations and risks, and would therefore require different risk thresholds and context-specific risk management approaches. For instance, using AI to improve the user experience for video games would differ considerably from a use case where AI assists hiring managers in selecting potential candidates. To ensure coherence with our national position on AI governance, the Government will also establish a common platform for regulatory agencies to coordinate on AI developments in their sectors, and share best practices when governing AI.

-	Consider updates to broader standards and laws to support effective AI use.

-	Contribute actively to international discourse on AI governance, to raise capacity, share best practices, and shape rules around AI, together with the international community.

Action 14

Raise security and resilience baseline for AI

The Government will also elevate the security and resilience baseline for all system owners using AI.

-	In the short term, we will work with partners to update cybersecurity toolkits for enterprises and individual users, to address AI-related risks. These include advisories and actionable guidelines on how system owners can augment their cybersecurity foundations to enable secure AI adoption.

-	In the longer term, we will also coordinate efforts to share best practices and ensure alignment to improve AI security, including through the development of standards and solutions for AI security. This will involve working with the private sector to co-create technical guidelines and solutions to secure AI use cases, and to leverage crowdsourcing initiatives to evaluate risks to AI (e.g. joint red teams to test AI systems). Such close partnerships with the private sector, including on capacity building and public education programmes, will raise baseline AI security capabilities across the ecosystem.

 

ENABLER 10: Leader in Thought and Action

Today’s international AI landscape is contested and fragmented.

-	AI has become a major front of competition. Global powers like the US and China dominate the global AI landscape in research outcomes, innovation capacity, and talent networks. In turn, critical and emerging technologies like AI have become a pre-eminent domain for geostrategic contestation.

-	While the international community is working to develop consensus around the responsible development and deployment of AI, efforts remain nascent at this point. All countries recognise the importance of managing AI well, and several have offered ideas on how the international community might cooperate on this. It is still early days, and it will take some time before the international community converges around any set of agreed principles.

Nevertheless, it remains critical for Singapore to stay engaged internationally, in a multi-party, multi-stakeholder fashion, to work towards a common understanding on AI.

-	AI is a technology “without a passport”. Its impact cannot be easily contained within any single country. It is therefore crucial for all countries to converge on efforts to make AI systems safer, and to avoid AI creating strategic risks and instability. This will require all countries to set aside the instincts of protectionism, to steer AI as a force for good. Singapore is determined to play our part, and we will continue to participate constructively at international fora on AI.

-	Such conversations on the trajectory of AI must be inclusive. AI will affect and potentially disrupt all of us. It is therefore important to widen the conversation, beyond the few who are developing cutting-edge models, to take in views from other countries and stakeholders, and build up the capacity for meaningful efforts to address AI’s impact.

-	Industry also plays a critical role to shape more responsible AI. We welcome ongoing industry-led efforts, including those by leading AI companies, as valuable contributions to the broader conversation on how to govern AI models better. We look forward to learning together with them.

Action 15

Establish Singapore as an ambitious and pragmatic international partner on AI innovation and governance

Singapore is committed to being a serious and reliable international partner on AI. This is important for sustaining our international position as a trusted hub. We will:

-	Continue to grow Singapore’s international networks with key partner countries and leading AI companies.

-	Increase international mindshare in practical and risk-based approaches to AI. Singapore has gained global attention because of early efforts to integrate AI in key sectors such as finance, trade, and healthcare, as well as pragmatic Government tools such as Pair. We have developed governance frameworks and open-source testing toolkits such as AI Verify. These demonstrate Singapore’s commitment to developing and deploying AI well, and we will leverage these gains to be a global pace-setter at the forefront of AI.

We will contribute to international AI developments by:

-	Anchoring key bilateral relationships with selected partners from government and industry, through substantive initiatives and technical cooperation. These will allow Singapore to “start small and move quickly”, to establish common ground with like-minded partners, as pathfinders to broad-based multilateral cooperation.

-	Demonstrating alignment with key international fora and supporting worthwhile platforms. Singapore will support and actively participate in substantive multilateral, multi-stakeholder, or plurilateral initiatives, that seek to achieve an inclusive, practical, and rules-based global environment for AI.

-	Sharing Singapore’s experience and curating meaningful partnerships for capacity building. We will actively profile Singapore’s approaches to AI through public engagements and conferences such as Asia Tech x Singapore (ATxSG) and the Singapore Conference on AI. We are the convenor of the Forum of Small States (FOSS), which now has a digital pillar of engagement. Together with government and industry partners, we will develop AI-related capacity building initiatives to benefit the 108 members of FOSS.

 

[Concluding remarks omitted.]