Singapore: A Bridge for Global AI

In its 60 year history, Singapore has proactively pivoted its whole economy multiple times.

Today, it recognizes that its open economy powered by a highly skilled workforce will be among the first to feel the full impact of advanced AI, both its upsides and challenges.

“Sometimes, it is the more vulnerable countries that are the first to want to institute guardrails, and the major powers eventually come on board.” - President Tharman Shanmugaratnam

The Singapore “AI pivot” has begun, and AI safety is a key priority. Singapore’s thesis is that those that deploy AI safely and responsibly will be the ones who reap the most benefits. It is also taking a proactive role in facilitating global AI coordination because as a small state plugged into the global economy, it has an existential stake in getting this right.

The 2026 budget announced a National AI Council to be chaired by Prime Minister Lawrence Wong whose responsibility will be to drive this transition.

“Our advantage lies in deploying AI effectively, responsibly, and at speed. Singapore can be a trusted hub where companies and researchers come together to develop, test and deploy impactful AI solutions – and do so faster and more coherently than many larger countries,” said PM Wong.

We believe that Singapore is uniquely positioned to play a global leadership role in AI safety. At Singapore AI Safety Hub (SASH), we have been fortunate to witness and contribute to this growth. In this blogpost, we share our reflections and observations on recent AI safety activities in Singapore, and opportunities to get involved.

Singapore’s case for a global AI safety role isn’t new. Its reputation as a trusted convenor is grounded in a track record: its support for UNCLOS helped states with sharply divergent interests establish shared rules for a critical global commons. AI safety presents a structurally similar challenge: technically complex, economically consequential, and deeply geopolitically sensitive. Progress depends on neutral parties that can bridge technical, policy, and diplomatic communities. Singapore’s pragmatic governance culture, strong state technical capacity, and trusted relationships are well positioned to shape practical AI safety norms, coordination mechanisms and institutional architectures for the global good.

We invite the international AI safety community to look at Singapore not just as a participant in the AI safety landscape, but as a platform to accelerate shared AI Safety priorities globally.

Singapore’s strengths in AI Safety

World-leading AI Safety Institute. In 2024, IMDA established the Singapore AI Safety Institute (Singapore AISI) to conduct safety and alignment testing for high-impact and generative AI models. It is one of the ten founding members of the international network of AISIs, which collaborate and share know-how on safe AI. The AISIs of US, UK and Singapore led the first joint testing exercise for AI models in the international network of AISIs. Singapore also co-led a joint testing exercise to determine whether guardrails on LLMs work in languages other than English. This has made Singapore AISI stand out as a competent and accessible international partner for the AI governance ecosystem.

Practical deployment of AI safety frameworks. Singapore deploys its AI safety research findings through voluntary testing tools and frameworks through the AI Verify Foundation. In January 2026, the AI Verify foundation released a comprehensive toolkit providing step-by-step guidance for pre-deployment LLM safety testing covering hallucination, bias, undesirable content, data leakage, and adversarial prompt vulnerabilities. Other initiatives include the Model AI Governance Framework for Generative AI (2024), AI Verify — a testing platform aligned with eleven global principles including transparency, robustness, fairness, and accountability — and Project Moonshot for LLM evaluation. These are actual codebases with datasets that companies can use to test their AI systems and generate compliance reports across multiple jurisdictions.

Active policy engagement with industry. Singapore serves as the regional headquarters for many multinational corporations, and it has actively engaged them in AI assurance just as in other policy matters in the past. The AI Verify Foundation has grown to over 120 member organizations — including AWS, Google, Microsoft, Meta, and Anthropic. The Global AI Assurance Pilot, launched in February 2025, pairs AI testing vendors with organizations deploying real-world GenAI applications to establish best practices that can inform future standards. During the recent Singapore AI Research Week, the AI Verify Foundation organized the AI Assurance Base Camp, a practitioner-focused event designed to facilitate adoption of these testing tools and share implementation best practices, and a Red-Teaming challenge. AI players like AI Singapore, Anthropic, AWS, Cohere, Google, and Meta, and experts in fields like sociology, cultural studies, and linguistics came together to test models to see if they generate harmful content against any social groups of Southeast Asia.

Regular host of international convenings. In January 2026, Singapore hosted the Singapore AI research week, with a busy calendar of activities, including the AAAI conference. This follows the success from last year when the AI research week included the ICLR conference, and the pivotal Singapore Conference on AI which brought “godfather of AI” Yoshua Bengio and other AI scientists into conversation about the future of AI and its safety. The Singapore Conference on AI produced the Singapore Consensus on Global AI Safety Research Priorities, a document that identifies and synthesises research priorities in AI safety. Such events are enthusiastically attended by government agencies such as the Ministry of Digital Development and Information, led by Josephine Teo, who has been an advocate for AI and AI safety activities at local and international conferences.

Strong technical research activities. At least 14 research groups across Singapore’s world-class universities are actively engaged in technical AI safety research. Particular focus areas following grants and local research priorities include robustness of foundation models, techniques for mitigating “supercharged disinformation”, knowledge-editing and agentic safety. Governmental research organisations have produced content moderation models and general purpose open-source LLMs for Singaporean and South East Asian contexts. Singapore’s GovTech has also produced AI Guardian - a platform for safety testing and governance which is used to test government AI services before deployment. We are seeing how public research in Singapore proactively identifies local AI Safety research priorities and advances them.

Substantial research funding expansion. Singapore is investing $1 billion over the next five years specifically into public AI research. This is a large investment for a city state of just 6 million residents. This funding will grow the AI safety presence in public research institutions, with explicit focus on resource-efficient AI, responsible AI, emerging AI methodologies, and general-purpose AI. Singapore’s AI Safety Institute has also awarded research grants on topics relevant to its priorities. One can expect relevant academic research in AI Safety to find funding in Singapore.

Growing AI Safety community. The Singapore AI Safety Hub has organised 30+ events in its first year, bringing together a total of 1000+ attendees. It has an active coworking space hosting professionals working in AI safety, and supporting career transitions and new talent entering the field. Some of the co-working members in SASH work in AI Safety organisations such as FAR.AI, Concordia AI, TruthfulAI, Apart Research, The Future Society. AI Safety organisations such as FAR.AI and Concordia AI are in the midst of expanding their Singapore presence. Third-party AI assurance providers like Resaro.ai and AIDX also have emerged. We see the AI safety ecosystem in Singapore continuing to grow and diversify for the foreseeable future.

Civil servants partner with AI safety experts. Through communications and working sessions with government agencies including IMDA, CSA, GovTech and MDDI, we’ve observed genuine curiosity about frontier AI safety topics. The engagement shows civil servants are looking to deepen their understanding of technical AI safety, particularly on distinguishing between commercial AI assurance and existential risk considerations, and evaluating which international research directions matter most for Singapore’s context. Singapore’s main strategic interests in AI are: steering AI for the public good, maintaining economic competitiveness as AI advances, and ensuring potential abuse and mismanagement risks are addressed. The stable political situation in Singapore allows for planning and implementation of strategy over decades rather than election cycles. Thus AI safety experts voicing thoughtful new perspectives can expect to find their views carefully considered and incorporated into national planning.

Local constraints

Singapore’s strengths in AI safety lie less in frontier model-building and more in shaping how advanced AI is governed, deployed and coordinated across borders. Singapore has a smaller domestic frontier AI safety research base than the US, the UK, and China, as well as limited frontier-model development. This has led to a deliberate strategic focus on governance innovation, deployment practices and international coordination — areas where Singapore’s influence can scale well beyond its size.

Singapore’s approach also leans toward voluntary and incentive-driven frameworks that work with industry. As a result, Singapore’s influence flows through persuasion, practical utility and convening power, not unilateral regulation. Understanding this helps calibrate expectations about what engagement here can realistically achieve.

Organizations that show up with substance, maintain consistent presence, and work constructively across government and industry are increasingly invited into deeper, more consequential conversations. This is what we’ve experienced at SASH over the past year, and it shapes our outlook on what’s possible.

Our outlook is that Singapore will continue its proactive and consultative approach — and move as fast as credible partners engage with them.

What this means for you

If you’re in industry or technical research: Singapore offers something unique - government partners who could support and facilitate AI safety research and development projects while involving you in policy consultations. Projects that demonstrate regional relevance (like AI-accelerated cybercrime in SEA) receive especially strong reception. We ourselves are actively working with local and international partners on projects related to AI Agent governance and verification mechanisms and are looking for experienced collaborators.

If you are working on international governance and coordination: Singapore remains an ideal location for such work. Its location, infrastructure and visa policies as well as its multiculturalism have led to multinational companies (Google DeepMind, OpenAI, ByteDance, etc.) opening regional offices in Singapore. At the same time, government backed initiatives such as the Singapore Consensus on AI Safety Research Priorities have created space for international dialogue. Singapore’s role in international convenings has always included the most important geopolitical matters such as discussed at the annual Shangri-La Dialogue, and these convenings will only feature more discussions on AI in the years to come. At SASH we’ve leaned into this by regularly collaborating with partners to organise workshops that bring together researchers from around the world to exchange views and build consensus.

If you’re focused on AI governance: Singapore’s voluntary framework approach means multiple stakeholder groups have influence. The focus on operationalization (shipping actual tools) means concrete technical contributions carry more weight than position papers.

How we can help

SASH operates at the intersection of these dynamics. We run a co-working space in central Singapore, host a wide range of AI Safety programmes, and lead both technical and governance focused AI research projects between local and international collaborators. We are co-located with Lorong AI, a Singapore government initiative which brings together government, industry, and research to build Singapore’s AI ecosystem. This is a collaborative hub where AI practitioners connect, innovate, and shape the future of AI in Singapore. In numbers, Lorong AI has 250+ members, 8500+ in their community network and ran 150+ events in 2025.

How we can help:

• Briefings on Singapore’s AI governance landscape, regulatory environment, and stakeholder priorities.
• Advising on stakeholder engagement across Singapore’s government, industry, and research sectors.
• Fellowship opportunities and capacity-building initiatives for technical AI safety work in the region.
• Office space access and event hosting for researchers and practitioners working on AI safety and governance in Southeast Asia.

We’re not the only way to engage with Singapore’s AI safety ecosystem, but we’ve spent the past year working closely across government, industry, and research institutions to build a deep, practical understanding of how this ecosystem functions. If you’re exploring how Singapore’s infrastructure could advance your work, we’re happy to share what we’ve learned.

Interested in exploring collaboration or opportunities to work on advancing AI safety and governance via Singapore? Reach out to us at hello@aisafety.sg.

Authors: Shashvat Shukla, Red Bermejo, Miro Plueckebaum, Valerie Pang

Singapore AI Safety Hub is building Singapore’s AI safety ecosystem, positioning it as a bridge for international AI safety coordination. We engage government and civil servants through briefings and our co-working space co-located with Lorong AI, run fellowship programs to develop talent, seed research initiatives, and coordinate AI safety activities across the island.

This analysis draws on insights from Concordia AI’s “State of AI Safety in Singapore (July 2025)” and our working experience engaging with Singapore’s AI governance ecosystem.

Further Resources

State of AI Safety in Singapore (July 2025). Concordia AI https://concordia-ai.com/research/state-of-ai-safety-in-singapore/

Building Trustworthy AI. AI Verify Foundation.

https://aiverifyfoundation.sg/

Model Governance Framework for Generative AI. https://aiverifyfoundation.sg/wp-content/uploads/2024/05/Model-AI-Governance-Framework-for-Generative-AI-May-2024-1-1.pdf

Singapore’s Approach to AI Governance. Personal Data Protection Commission Singapore. https://www.pdpc.gov.sg/help-and-resources/2020/01/model-ai-governance-framework

Governing AI: Singapore’s Dynamic Approach. Civil Service College Singapore. https://knowledge.csc.gov.sg/governing-ai-singapore-s-dynamic-approach/

AI Watch: Global regulatory tracker - Singapore. White & Case. https://www.whitecase.com/insight-our-thnking/ai-iwatch-global-regulatory-tracker-singapore