Introduction
Singapore’s copyright law is among the friendliest in Asia to AI developers. The government now wants to know whether that law, and its patent regime, still strike the right balance. On 26 August the Ministry of Law (MinLaw) and the Intellectual Property Office of Singapore (IPOS) opened a public consultation on how AI affects the country’s intellectual property framework. It closes at 5pm on 22 October. AI developers, rights holders and companies deploying AI tools have just over two weeks to comment.
What is on the table
The consultation does not propose any law changes yet. It sets out the questions officials think matter most, after earlier talks with technology firms, creative industries, lawyers and academics.
| Area | Key questions |
|---|---|
| AI training (copyright) | Is the scope of the computational data analysis exception clear enough? How should “lawful access” and rights-owner safeguards work in practice? Should the Government support voluntary, industry-led technical measures, such as opt-out signals? |
| AI outputs (copyright) | How do existing principles apply when AI output infringes? How should responsibility be divided among AI developers, deployers and end users? |
| Human creativity (copyright) | How should human contribution be understood, recognised and evidenced when generative AI is used as a tool? |
| Inventorship (patents) | How do inventorship rules apply when humans frame the problem, choose among AI outputs or modify AI-generated solutions? |
| Prior art (patents) | How will mass publication of AI-generated technical disclosures affect searches, examination and incentives to innovate? |
Why it matters
Singapore’s Copyright Act 2021 already includes an exception for computational data analysis, which covers using works to train AI, provided the user has lawful access to the material. Together with Japan, Singapore has been seen as one of the more permissive places for AI training in the region. The consultation suggests officials are now worried that the exception is not clear enough. In particular, they want to know how lawful access is shown, and whether rights owners need better ways to signal or enforce their preferences.
The questions about outputs are probably the more commercially sensitive. Most companies now use AI built by others. If AI-generated content infringes copyright, the split of liability between model developer, the business that deploys it and the person who prompts it will shape procurement contracts, indemnities and insurance across the region.
On patents, most major patent systems treat only natural persons as inventors. The consultation asks how much human involvement is enough when AI does much of the inventive work. That matters to every company using AI in research and development.
How to respond
Submissions can be made by FormSG (separate forms for copyright and patents) or by email to MinLaw. Officials may publish submissions and name the source. Confidential material must be clearly marked and placed in a separate annex, and a request to keep all or most of a submission confidential will not be accepted.
For practitioners, the consultation is also a guide to current thinking. Even without draft legislation, it shows where Singapore is likely to move. Companies should review now how they document training data sources, how they allocate output risk in vendor contracts, and how they record human contribution to AI-assisted work.
Summary
MinLaw and IPOS are asking whether Singapore’s IP regime still fits the realities of AI, across training, outputs, authorship, inventorship and prior art. No law changes are proposed yet, but the questions show where policy is heading: more certainty on training, clearer allocation of liability for outputs, and closer attention to human contribution. The deadline is 22 October. Organisations that build or rely on AI should consider responding, and should check their own contracts and record-keeping against these questions.
Leave a Reply