Technology

AI Rules in Practice: Singapore’s Position

13 February 20265 min read

Introduction

Imagine a world where machines can think, learn, and make decisions like humans. That's the promise of artificial intelligence (AI). From self-driving cars to chatbots that answer your questions, AI is changing how we live and work. But with great power comes great responsibility. What if an AI system makes a mistake that hurts someone? Who pays for it? How do we ensure AI is fair and safe?

Unlike the European Union, which has opted for sweeping legislation like the AI Act^[1] that categorises AI systems by risk level and imposes strict compliance requirements, Singapore has chosen a markedly different path. The city-state has positioned itself as the pragmatic alternative in the global AI governance debate, favouring voluntary frameworks, industry collaboration, and what Deputy Prime Minister Gan Kim Yong describes as a "nimble" and "flexible—but at the same time, very careful" approach to providing guidance and guardrails.^[2]

What Does "AI" Mean in Singapore?

First things first: What is AI? In Singapore, it's defined in a user-friendly way through the Model Artificial Intelligence Governance Framework, issued by the Infocomm Media Development Authority (“IMDA”) and the Personal Data Protection Commission (“PDPC”). AI refers to technologies that simulate human traits like knowledge, reasoning, problem-solving, perception, learning, and planning, producing outputs such as predictions, recommendations, or classifications.^[3]

Singapore distinguishes between two main types:

Traditional (Discriminative) AI: Systems that analyse historical data to make predictions, like a bank's credit scoring tool.
Generative AI: Systems that create new content, such as text, images, or code, like ChatGPT.

Crucially, Singapore's definition appears in a voluntary framework, not a statute. This distinction reveals everything about Singapore's philosophy: guide first, regulate later if necessary.

The Model AI Governance Framework

Launched in 2019 and updated in 2020, the Model AI Governance Framework applies to traditional AI. It is built on two core principles^[4]:

Explainability, Transparency, and Fairness: Organizations must explain AI decisions, operate transparently, and ensure fairness.
Human-Centricity: AI should prioritize human well-being and safety.

The framework covers four areas: internal governance, human involvement in decisions, operational management (including data quality), and stakeholder communication.

The Model Governance Framework for Generative AI (2024)

With the rise of tools like ChatGPT, Singapore introduced the Model Governance Framework for Generative AI in June 2024^[5]. It outlines nine dimensions for building trustworthy systems:

Accountability
Data quality and availability
Trusted development and deployment
Incident reporting
Testing and assurance
Security
Content provenance (tracking origins of AI-generated content)
Safety and alignment with human values
Global cooperation

Singapore's National AI Strategy 2.0

In December 2023, Singapore launched its updated National AI Strategy^[6] (“NAIS 2.0”), building on the 2019 version. It makes three key shifts:

AI as a "necessity," not just an opportunity—everyone must understand it.
A global focus: Connect to international networks, contribute breakthroughs, and create valued AI products.
From Projects to Systems: Scale up from pilots in healthcare, education, and security to widespread implementation.

NAIS 2.0 emphasises regulatory agility, using a mix of rules and guidelines in coordination with global partners.

The Surprising Truth: No Comprehensive AI Law

Unlike the EU's AI Act, Singapore has no overarching legislation specifically governing artificial intelligence across all sectors. The government explicitly states it is not currently looking to enact broad AI regulations, preferring to monitor technological developments before deciding on legislative approaches.^[7]

This isn't regulatory apathy-it's deliberate strategic patience. Singapore's approach rests on three pillars: targeted sectoral rules where safety demands them, voluntary frameworks for general guidance, and practical tools that help organisations self-regulate.

Where hard rules do exist, they address specific high-risk applications:

Autonomous Vehicles: The Road Traffic Act was amended in 2017 to allow self-driving car trials, replacing a system based on exemptions. Operators must carry liability insurance covering at least SGD $1.5 million.
Medical Devices: AI-powered medical devices must register under the Health Products Act, with additional requirements for training data documentation and performance validation.
Election Integrity: In September 2024, Singapore introduced its first law explicitly mentioning AI—the Elections (Integrity of Online Advertising) Amendment Bill. Effective January 2025, it prohibits publishing AI-generated deepfake content depicting election candidates saying or doing things they never actually said or did^[8]. The law carries penalties of up to SGD $1,000 in fines or 12 months imprisonment, with platforms facing up to SGD $1 million for non-compliance.

Who is Responsible When AI Fails?

One of the biggest hurdles in AI law is the "Black Box" problem: if an AI makes a mistake, it’s often impossible to explain why^[9].

In Singapore, there are no special "AI liability" laws yet. Instead, we rely on:

Contract Law: If a business's AI software fails, the contract between the developer and the user usually dictates who pays.
Tort Law (Negligence): A victim must prove the developer or operator owed them a "duty of care" and failed to meet a reasonable standard.
Criminal Law: AI cannot be "arrested." Criminal liability stays with the humans—the programmer, owner, or operator—depending on their intent or knowledge.

Intellectual Property: Can AI be an "Inventor"?

For those looking to patent AI creations, the answer in Singapore is currently no.

Patents: Singapore law requires an inventor to be a "natural person." A machine cannot own a patent.
Copyright: For an image or text to be copyrighted, there must be "human intellect" involved. "AI-assisted" works (where a human uses AI as a tool) may be protected, but "AI-generated" works with no human intervention generally are not.

AI in the Workplace and Privacy

Businesses using AI face two major hurdles:

Workplace Fairness: While AI is used for hiring and monitoring, the Workplace Fairness Act (passed Jan 2025) ensures these tools cannot be used to discriminate based on gender, age, or disability.
Data Privacy (“PDPA”): The Personal Data Protection Act applies to AI. Companies must be transparent about how they use personal data to train models and must ensure that data is kept secure.

Conclusion

Singapore treats AI like a powerful tool. By emphasising strategies, voluntary frameworks, and targeted rules, it fosters trust without stifling growth. For everyday folks, this means safer AI in healthcare, transportation, and the workplace. Businesses get guidance to innovate responsibly. Privacy and fairness are priorities, with evolving tools to handle data and liability. As AI evolves, Singapore's agile approach-watching, consulting, adapting-positions it as a global leader. Whether you're a user, developer, or curious citizen, understanding these rules helps navigate the AI future. It's not about fearing machines; it's about making them work for us.

Need legal guidance?

Our team can help you navigate these legal matters with clarity and confidence.

Get in Touch More Articles