🇸🇬Singapore

Overview

Singapore’s regulatory philosophy regarding artificial intelligence is defined by a pragmatic, "pro-innovation" stance that seeks to position the city-state as a global AI hub while ensuring robust safeguards for public trust. This dual-track strategy is encapsulated in the National Artificial Intelligence Strategy 2.0 (NAIS 2.0), launched in late 2023, which transitions the nation from a project-based AI adoption model to a systems-wide integration. The Singaporean government views AI not as an optional enhancement but as a necessity for national competitiveness and social well-being. Consequently, the regulatory environment is designed to be agile, responding to technological shifts—such as the rapid emergence of generative AI—without imposing the kind of rigid, horizontal legislative burdens seen in other jurisdictions. By fostering a "trusted AI" ecosystem, Singapore aims to attract global talent and investment while ensuring that AI applications are safe, fair, and human-centric. This approach is deeply integrated into the broader "Smart Nation" initiative, which envisions a digital-first society where technology serves the collective good. The government acts as both a regulator and a catalyst, funding research into AI safety while simultaneously setting the boundaries for its ethical use. This unique positioning allows Singapore to move faster than many of its peers, implementing practical solutions like the AI Verify testing framework to provide tangible evidence of AI performance and safety. The maturity of Singapore's AI landscape is reflected in its shift from high-level ethical principles to practical, implementable toolkits and targeted statutes. While the foundational Model AI Governance Framework (first issued in 2019) remains a cornerstone of the voluntary compliance regime, recent years have seen the introduction of binding laws to address specific societal harms. These include the Elections (Integrity of Online Advertising) (Amendment) Act 2024, which targets deepfakes in political contexts, and the upcoming Workplace Fairness legislation. This evolution demonstrates a sophisticated understanding of risk: maintaining a light touch for general-purpose innovation while applying hard-law interventions where AI-driven outcomes could fundamentally undermine democratic integrity, workplace equity, or online safety. This balanced approach ensures that Singapore remains at the forefront of global AI governance dialogue, often serving as a bridge between the prescriptive models of the West and the more laissez-faire environments elsewhere. The focus is consistently on "human-in-the-loop" or "human-over-the-loop" configurations, ensuring that even as systems become more autonomous, accountability remains firmly with human operators and organizations.

Regulatory approach

Singapore’s regulatory approach is characterized by its sector-led, risk-based, and multi-stakeholder nature. Rather than adopting a single, overarching "AI Act," Singapore leverages its existing legal architecture—such as data protection, consumer rights, and sector-specific regulations—supplemented by agile, non-binding frameworks. This horizontal-sectoral hybrid model allows agencies like the Monetary Authority of Singapore (MAS) and the Ministry of Health (MOH) to tailor requirements to the unique risks of their respective domains. For instance, the MAS FEAT principles provide specific guidance for the financial sector, while the PDPC’s Advisory Guidelines offer cross-sectoral clarity on personal data usage. This ensures that regulation is proportionate to the risk of harm, with high-stakes applications in finance or healthcare receiving more intensive oversight than low-risk consumer recommendation engines. The government also utilizes regulatory sandboxes, such as the IMDA’s GenAI Sandbox, to allow companies to test innovative AI solutions under supervised conditions, providing regulators with real-time data on emerging risks before formal rules are finalized. The distinction between binding "hard law" and non-binding "soft law" is a deliberate feature of the Singaporean model. The government frequently utilizes "Registrar’s Circulars," "Advisory Guidelines," and "Model Frameworks" to set supervisory expectations. These instruments are technically voluntary but carry significant weight; non-compliance can signal poor corporate governance to regulators and may lead to increased scrutiny under existing statutes like the Personal Data Protection Act (PDPA). However, when voluntary measures are deemed insufficient to protect the public interest, the government pivots to legislative amendments. The Online Safety (Relief and Accountability) Bill (OSRA) is a prime example, creating statutory causes of action and administrative powers to compel the removal of harmful AI-generated content. This flexibility allows Singapore to maintain a stable yet responsive regulatory environment that evolves alongside the technology it governs. Furthermore, the government actively engages with the private sector through the AI Verify Foundation, a non-profit consortium that brings together global tech leaders to develop open-source testing tools, ensuring that Singapore's standards are both technically feasible and commercially viable. The governance of AI in Singapore is a distributed responsibility involving several key agencies coordinated under the "Smart Nation" umbrella. The Infocomm Media Development Authority (IMDA) is the lead agency for industry development and technical standards. It is responsible for stewarding initiatives like the AI Verify Foundation and the Model AI Governance Framework for Generative AI. The IMDA's role is unique in that it balances the promotion of the AI industry with the enforcement of online safety standards. Working alongside it, the Personal Data Protection Commission (PDPC) ensures that AI deployments comply with the PDPA. The PDPC provides the necessary oversight for data-driven algorithmic systems, issuing advisory guidelines that clarify how existing data protection principles apply to complex AI architectures. These agencies often collaborate to produce joint guidelines, ensuring a unified government voice on issues like synthetic data and AI-driven recommendation engines. The Smart Nation Group, housed within the Prime Minister's Office, provides the high-level strategic direction and cross-agency coordination required for the National AI Strategy, ensuring that AI policy is aligned with national economic and social goals. In addition to these central bodies, sectoral regulators play a critical role in enforcement and oversight within their jurisdictions. The Monetary Authority of Singapore (MAS) is particularly active, overseeing the responsible use of AI in the financial sector through the FEAT principles and the Veritas initiative. The MAS conducts thematic reviews of financial institutions' AI risk management practices and has the power to issue supervisory directions to ensure financial stability and consumer protection. For online safety, the newly proposed Office of the Commissioner of Online Safety (under the OSRA Bill) will have the power to investigate reports and issue remedial directions to platforms, including content removal and account restriction orders. This decentralized model ensures that regulators with deep domain expertise are responsible for the AI applications most relevant to their sectors, while the IMDA and PDPC provide the foundational governance standards that apply across the entire economy. This approach prevents regulatory gaps and ensures that AI oversight is as nuanced as the industries it affects.

Enforcement & penalties

Enforcement in Singapore’s AI landscape follows a graduated approach, ranging from administrative directions to substantial financial penalties. Under the Personal Data Protection Act (PDPA), the PDPC can impose fines of up to S$1 million or 10% of an organization’s annual turnover in Singapore, whichever is higher, for serious data breaches involving AI systems. These penalties are designed to be a significant deterrent, particularly for large multinational corporations. The Elections (Integrity of Online Advertising) (Amendment) Act 2024 introduces even more significant deterrents for platforms, with social media companies facing fines of up to S$1 million for failing to comply with corrective directions regarding manipulated election content. These financial sanctions are often accompanied by remedial directions, such as requirements to cease data processing, delete illegally obtained datasets, or implement specific technical safeguards like bias mitigation audits. The government’s philosophy is that enforcement should be proportionate to the harm caused, with a focus on remediation and systemic improvement rather than purely punitive measures. Beyond financial penalties, the legal framework provides for criminal sanctions and civil remedies in specific contexts. The OSRA Bill, for instance, enables the Online Safety Commission to issue directions for content removal and account restriction, with non-compliance potentially resulting in criminal prosecution for platform executives. In the judicial sphere, the Supreme Court’s "Guide on the Use of Generative AI Tools" warns that lawyers or litigants who submit AI-generated fabrications may face disciplinary action, costs orders, or contempt of court proceedings. This enforcement regime is supported by robust appeals processes, typically involving internal reconsideration by the relevant commissioner followed by the right to appeal to specialized tribunals or the High Court, ensuring procedural fairness for all stakeholders. The Online Criminal Harms Act (OCHA) also provides for the blocking of online services that fail to comply with directions, a "nuclear option" that underscores the government's commitment to maintaining a safe digital environment. This multi-faceted enforcement strategy ensures that there are clear consequences for the misuse of AI, whether by individuals, corporations, or global tech platforms.

Data protection

The Personal Data Protection Act (PDPA) serves as the fundamental framework for the collection, use, and disclosure of personal data in Singapore’s AI ecosystem. Organizations developing or deploying AI must adhere to the PDPA’s core obligations, including the Accountability Obligation, which requires firms to demonstrate compliance through documented policies and impact assessments. The PDPC’s 2024 "Advisory Guidelines on the Use of Personal Data in AI Recommendation and Decision Systems" clarify that while consent remains the primary lawful basis for processing, organizations may rely on exceptions such as the "Business Improvement Exception" for AI model development. This exception allows companies to use personal data without consent for purposes like improving products or services, provided they meet strict criteria regarding data minimisation and can demonstrate that the purpose cannot be achieved through anonymized data. This provides a vital pathway for AI innovation while maintaining a high bar for privacy protection. Singapore also places a strong emphasis on Privacy Enhancing Technologies (PETs) and the use of synthetic data to mitigate privacy risks. The PDPC’s "Proposed Guide on Synthetic Data Generation" provides a structured five-step approach for organizations to create artificial datasets that retain the statistical utility of real data while protecting individual identities. This focus on PETs is supported by the IMDA’s PET Sandbox, which allows companies to experiment with data-sharing models—such as federated learning or homomorphic encryption—in a controlled regulatory environment. By integrating these technical safeguards into the legal framework, Singapore ensures that its data protection regime remains robust enough to handle the complexities of large-scale AI training while remaining flexible enough to support data-driven innovation. The PDPC also emphasizes the "Notification Obligation," requiring organizations to be transparent with individuals about how their data is being used in AI systems, thereby fostering a culture of trust and informed participation in the digital economy.

Sector-specific rules

In the financial sector, the Monetary Authority of Singapore (MAS) has established a comprehensive suite of expectations centered on the FEAT Principles (Fairness, Ethics, Accountability, and Transparency). These principles are operationalized through the Veritas initiative, which provides financial institutions with technical toolkits to assess their AI models for bias and explainability. In late 2025, MAS further strengthened this regime by consulting on new "Guidelines on AI Risk Management," which propose mandatory AI inventories and materiality assessments for all financial institutions. These rules are particularly focused on high-stakes use cases such as credit decisioning, insurance underwriting, and autonomous "agentic" systems that execute financial transactions on behalf of customers. The MAS approach is highly collaborative, involving industry working groups to develop sector-specific benchmarks for fairness, ensuring that the rules are grounded in the practical realities of financial services. The legal and judicial sectors have also introduced targeted rules to manage the risks of generative AI. The Supreme Court of Singapore’s "Registrar's Circular No. 1 of 2024" mandates that all court users remain fully responsible for the accuracy of AI-generated submissions and must verify every legal citation against authoritative sources. This is a direct response to the risk of AI "hallucinations" in legal proceedings. Similarly, in the employment sector, the Workplace Fairness Bill prohibits the use of discriminatory algorithms in hiring and appraisals, requiring employers to maintain transparent grievance-handling processes. In the healthcare sector, the Ministry of Health (MOH) has issued guidelines for the use of AI in clinical decision support systems, emphasizing the need for clinical validation and human oversight. These sector-specific interventions ensure that the unique risks of AI—whether they be biased recruitment or inaccurate medical diagnoses—are addressed by the authorities best equipped to understand the nuances of those professional environments, preventing a "one-size-fits-all" approach that could stifle beneficial applications.

International alignment

Singapore is a proactive participant in global AI governance, emphasizing the need for international interoperability and the alignment of standards. The IMDA’s AI Verify toolkit is a key instrument in this effort, designed to map to international frameworks such as the NIST AI Risk Management Framework and ISO/IEC standards. By creating a testing regime that is recognized globally, Singapore helps its local firms navigate foreign regulatory environments while ensuring that international AI providers can enter the Singaporean market with confidence. Singapore also plays a leading role in regional initiatives, notably the development of the "ASEAN Guide on AI Governance and Ethics," which mirrors many of the principles found in Singapore's own Model Framework. This regional leadership is crucial for creating a harmonized digital market across Southeast Asia, reducing compliance costs for businesses operating across borders. While Singapore has not adopted the prescriptive, classification-heavy approach of the EU AI Act, it maintains a close dialogue with European regulators to ensure that its "soft law" toolkits can produce the documentation and evidence required for compliance in the EU. This "interoperability-first" strategy is further evidenced by Singapore’s participation in the Global Partnership on AI (GPAI) and its commitment to the OECD Principles on Artificial Intelligence. Through bilateral agreements and joint pilots—such as the Global AI Assurance Sandbox—Singapore seeks to harmonize the technical benchmarks used for AI safety and robustness. This ensures that its regulatory environment remains a bridge between different global governance philosophies, advocating for a model that is both ethically sound and commercially practical. By contributing to international standards, Singapore ensures that its voice is heard in the global conversation on AI ethics, helping to shape a future where technology is governed by shared values of transparency and accountability.

What's next

The next two years will see the full operationalization of several major legislative and policy initiatives. The Online Safety (Relief and Accountability) Bill is expected to lead to the launch of the Online Safety Commission in 2026, which will significantly expand the government’s powers to regulate AI-generated content on social media and messaging platforms. This will include the power to issue "stop-communication" directions and "account-restriction" directions to combat the spread of harmful synthetic media. Simultaneously, the implementation of the Workplace Fairness Act will reach its target commencement in late 2027, requiring thousands of employers to audit their automated HR systems for compliance with new non-discrimination standards. These developments represent a significant hardening of the regulatory environment for specific, high-impact AI applications, moving from voluntary guidance to mandatory statutory requirements. On the policy front, the National AI Strategy 2.0 will continue to drive investments in sovereign AI capabilities, including the National Multimodal LLM Programme (NMLP). This initiative aims to develop base models that are culturally and linguistically attuned to the Southeast Asian context, reducing reliance on Western-centric models and ensuring that AI systems used in Singapore are representative of its diverse population. Furthermore, the MAS is expected to finalize its Guidelines on AI Risk Management for Financial Institutions by early 2026, which will likely set a new global benchmark for central bank oversight of AI. As generative AI continues to evolve into more autonomous "agentic" systems, Singapore’s regulators have signaled that they will remain agile, using their established sandboxes and consultative models to update guidelines in real-time. This proactive stance ensures that the city-state remains a safe harbor for responsible AI innovation, capable of capturing the benefits of the technology while mitigating its most complex risks.