Singapore AI Regulation Overview

Overview

Singapore has developed a mature, pragmatic, and multi-layered AI regulatory landscape designed to foster technological innovation while maintaining robust safety, ethical, and data protection standards. Under its overarching National Artificial Intelligence Strategy 2.0 (NAIS 2.0), launched in December 2023, the nation shifted from project-level pilots to systems-level scaling across three key systems and ten enablers. NAIS 2.0 sets twin goals of Excellence—building selective peaks of AI research and industrial capability—and Empowerment, raising AI literacy, trust, and operational capabilities across citizens and businesses nationwide.

Singapore’s framework balances ecosystem enablement with targeted legislative and regulatory oversight. National initiatives such as AI Singapore (established in 2017) drive foundational research, talent pipelines, and industry translation through programs like 100 Experiments (100E) and the S$70 million National Multimodal LLM Programme (NMLP). Rather than enacting an overarching omnibus AI statute, Singapore relies on sector-specific regulatory updates, advisory guidelines under data protection laws, voluntary testing toolkits, and targeted legislation addressing acute societal risks such as deepfakes, online harms, and workplace bias.

Regulatory Approach

Singapore’s approach to AI regulation is predominantly soft-law, risk-based, and sector-specific, supplemented by legally binding rules where high-stakes risks require statutory enforcement. The foundation of this model rests on voluntary frameworks co-developed by government agencies and industry, such as the Model AI Governance Framework series administered by the Infocomm Media Development Authority (IMDA) and the Personal Data Protection Commission (PDPC). These frameworks translate high-level ethical principles—such as fairness, transparency, explainability, and human accountability—into operational guidance for traditional AI, generative AI, and autonomous agentic AI systems.

To support compliance and verification without imposing prescriptive statutory barriers, Singapore emphasizes technical evaluation tools and sandboxes. IMDA and PDPC launched AI Verify, an open-source testing toolkit and governance framework, along with specialized modules like Project Moonshot for Large Language Models and the Global AI Assurance Sandbox. When binding interventions are necessary, lawmakers deploy targeted legislative amendments within existing statutory regimes, such as electoral integrity statutes, employment non-discrimination laws, judicial circulars, and financial sector supervisory expectations issued by the Monetary Authority of Singapore (MAS).

Key AI Legislation

• Elections (Integrity of Online Advertising) (Amendment) Act 2024 (ELIONA): Enacted to preserve electoral integrity during election periods, this statute prohibits the publication, sharing, or boosting of digitally generated or manipulated content (including deepfakes) that realistically depicts candidates saying or doing things they did not do. It empowers the Returning Officer to issue corrective directions to communicators, social media services, and ISPs.
• Workplace Fairness Bill (2025): Part of a two-part statutory framework creating Singapore's first standalone protection against workplace discrimination across hiring, performance appraisal, and dismissal. It prohibits adverse employment decisions based on protected characteristics (including age, nationality, sex, race, religion, and disability) and establishes a Commissioner for Workplace Fairness and Employment Claims Tribunals adjudication framework.
• Online Safety (Relief and Accountability) Bill (OSRA Bill): Under parliamentary review, this Bill establishes the Office of the Commissioner of Online Safety and the Online Safety Commission (OSC) to provide swift administrative relief against online harms, including intimate image abuse, online harassment, and inauthentic material abuse (deepfakes). It grants powers to issue takedown, account-restriction, and right-of-reply directions, and creates statutory torts for civil redress.
• Advisory Guidelines on Use of Personal Data in Generative AI (2026): Issued by the PDPC under the Personal Data Protection Act 2012 (PDPA), these guidelines clarify lawful bases, consent rules, data minimisation, and Data Protection Impact Assessments (DPIAs) across the lifecycle of generative AI development and deployment.
• Advisory Guidelines on the Use of Personal Data in AI Recommendation and Decision Systems (2024): PDPC guidance clarifying how PDPA requirements apply to AI systems processing personal data for recommendations and automated decision-making, elaborating on the Business Improvement Exception and Research Exception.
• Updated Model AI Governance Framework for Agentic AI (2026): IMDA framework addressing risk management, human accountability, technical controls, and end-user responsibility for autonomous AI agents capable of planning and dynamic action.
• Guide on the Use of Generative AI Tools by Court Users (2024): Supreme Court Registrar's Circular No. 1 of 2024 setting binding procedural requirements for lawyers, litigants, and witnesses using generative AI in court submissions and evidentiary materials across all Singapore Courts.

Governance & Enforcement Bodies

AI governance in Singapore is decentralized across sector regulators, policy offices, and enforcement commissions. Strategic policy guidance and digital vision are led by the Smart Nation Group (Smart Nation and Digital Government Office) within the Prime Minister's Office, working alongside the Ministry of Digital Development and Information (MDDI). The Advisory Council on the Ethical Use of AI and Data, convened in 2018, provides high-level stakeholder advice to government and business on emerging technology ethics.

Operational oversight and technical standard-setting are primarily carried out by IMDA and the PDPC. IMDA stewards industry AI frameworks, the AI Verify Foundation, and platform codes of practice. The PDPC enforces data protection compliance under the PDPA across AI lifecycles. Specialized sector bodies exercise direct authority over AI applications within their domains, including the Monetary Authority of Singapore (MAS) for financial institutions, the Returning Officer for electoral advertising, the Supreme Court Registrar for court proceedings, and the Ministry of Manpower (MOM) alongside the Commissioner for Workplace Fairness for employment practices.

Penalties & Enforcement

Enforcement mechanisms in Singapore depend on whether a breach involves mandatory statutory provisions or non-compliance with sector supervisory directions. Under the Elections (Integrity of Online Advertising) (Amendment) Act 2024 (ELIONA), failure by designated social media services to comply with a Returning Officer's corrective direction carries criminal fines up to S$1,000,000 upon conviction. Individual offenders face fines up to S$1,000, imprisonment for up to 12 months, or both. Candidates who knowingly make false or misleading declarations regarding manipulated content commit illegal practices under election law, leading to potential fines or vacation of elected seats.

Under the Workplace Fairness Bill, non-compliance with administrative directions or orders issued by the Commissioner for Workplace Fairness carries civil and criminal sanctions. Corporate bodies face civil court penalties up to S$50,000 for a first order and up to S$250,000 for subsequent orders, while failure to obey directions can attract fines up to S$5,000. Under the OSRA Bill, non-compliance with Online Safety Commission directions can result in administrative remedies, account suspensions, and statutory tort actions for civil damages or injunctions. In court proceedings, non-compliance with the Guide on Generative AI Tools can result in personal costs orders against legal practitioners, discounting of evidence, disciplinary referrals, or contempt of court proceedings. Non-binding advisory guidelines issued by PDPC and MAS do not carry direct statutory fines, but non-compliance exposes entities to administrative penalties and directions under underlying statutes such as the PDPA or financial licensing laws.

Data Protection Framework

Data protection in AI processing is governed by the Personal Data Protection Act 2012 (PDPA) and enforced by the PDPC. The PDPA applies to the collection, use, and disclosure of personal data across all stages of the AI lifecycle, from raw dataset curation and model training to inference and output management. Organisations relying on personal data for AI development must establish a valid legal basis—typically through meaningful consent or statutory exceptions such as the Business Improvement Exception and Research Exception, provided specific statutory criteria and data minimisation safeguards are satisfied.

PDPC advisory guidelines require organizations to implement Privacy by Design, conduct Data Protection Impact Assessments (DPIAs), ensure dataset accuracy, and maintain de-identification or pseudonymisation where feasible. In July 2024, the PDPC issued the Proposed Guide on Synthetic Data Generation, outlining a five-step process to manage re-identification risks when generating structured synthetic data as a Privacy Enhancing Technology (PET). The corpus contains no record of statutory data localization mandates under Singapore's general data protection framework.

Sector-Specific Rules

Sector-specific AI regulation is tailored to the distinct operational risks of individual industries. In the financial sector, the Monetary Authority of Singapore (MAS) published the FEAT Principles (Fairness, Ethics, Accountability, and Transparency) in 2018, operationalised through the industry Veritas initiative. In December 2024, MAS published an Information Paper on AI Model Risk Management following thematic bank reviews, instructing financial institutions to maintain centralized AI inventories, perform risk materiality assessments, validate models independently, and implement continuous post-deployment performance monitoring and kill-switch controls.

In the judicial and legal sector, the Singapore Judiciary implemented Registrar's Circular No. 1 of 2024 (effective October 2024), establishing that court users remain strictly accountable for all filed documents. It prohibits the fabrication of evidence using generative AI, requires mandatory fact-checking of legal citations against authoritative sources such as Singapore Statutes Online, and empowers judges to demand affidavits explaining AI usage and verification protocols. In employment, the Workplace Fairness Bill prohibits AI-driven or human bias in recruitment, appraisal, and retrenchment across protected characteristics. For healthcare and autonomous vehicles, while public research investments through AI Singapore address clinical and urban deployment challenges, the corpus holds no record of standalone statutory regulations specific to those sectors.

International Alignment

Singapore actively aligns its AI governance tools and frameworks with international standards to ensure cross-border interoperability and facilitate international trade. The AI Verify framework and toolkit were designed to map directly onto international principles established by the OECD, ISO/IEC, G7, and the European Union AI Act principles. Stewardship of AI Verify by the AI Verify Foundation involves active technical collaboration with global benchmarking bodies, including MLCommons.

Through initiatives such as the Global AI Assurance Pilot and the Global AI Assurance Sandbox, Singapore pairs international deployers with specialist testing entities to create standardized evaluation methodologies for traditional and generative AI. Rather than adopting the rigid binding classification model of the EU AI Act, Singapore incorporates international risk management norms into flexible, risk-calibrated soft-law frameworks that adapt dynamically as AI capability evolves.

Future Developments

Singapore’s AI regulatory roadmap includes several pending legislative and policy milestones. The Online Safety (Relief and Accountability) Bill (OSRA Bill), introduced to Parliament in October 2025, sets an operational target for the Office of the Commissioner of Online Safety and the Online Safety Commission (OSC) in 2026. Implementation will begin with five core harms before progressively expanding to cover inauthentic material abuse (deepfakes) and other emerging online harms.

Following parliamentary passage of the Workplace Fairness Bill and the companion Workplace Fairness (Dispute Resolution) Bill in 2025, the target commencement date for the complete Workplace Fairness Act regime is set for 31 December 2027, appointed by Gazette notification. Additionally, MAS signaled plans to translate its model risk management observations into formal supervisory AI risk management guidelines for the financial sector, while the PDPC continues to refine its guidance on synthetic data generation and Privacy Enhancing Technologies through the IMDA PET Sandbox.

Enforcement Bodies