Singapore launches Public Consultation on AI and Intellectual Property
On 26 August 2026, the Ministry of Law (“MinLaw”) and the Intellectual Property Office of Singapore (“IPOS”) launched a public consultation on Artificial Intelligence and Singapore's Intellectual Property Regime (the “Consultation”),[1] along with the publication of a consultation paper (“Consultation Paper”).[2] The Consultation seeks to clarify established legal principles while inviting views on how Singapore legislation should develop to address these issues. The Consultation period runs until 22 October 2026.
Background
Singapore’s broader AI policy seeks to support innovation while maintaining trust, security, privacy and legitimate commercial interests. In the intellectual property context, this balancing exercise is complicated by the need to preserve the rights and incentives that drive human creativity, innovation and investment. In particular, significant policy questions have been raised in relation to copyright and patents.
Copyright
A. Fostering Greater Certainty and Accountability in AI Training
The Consultation Paper recognises that AI training may not infringe copyright by virtue of to the computational data analysis (“CDA”) exception under Singapore’s Copyright Act. According to the Consultation Paper, the CDA exception was expressly intended to apply to machine learning for all types of technologies. including generative AI, permitting use of copyright works if the user (AI developer) has lawful access to the works amongst other requirements.
Furthermore, the Consultation Paper notes that:
- The CDA exception applies to both commercial and non-commercial contexts;
- Contractual terms that exclude the CDA exception are unenforceable;
- The use of technical measures that do not create legal obligations on users (e.g. the Robots Exclusion Protocol) does not, in itself, render an AI models' access to a work unlawful; and
- It remains uncertain if breach of laws that give rise to civil liability (e.g. a breach of contract) will render an AI system's access to a work unlawful.
The Consultation Paper sets out several illustrative examples demonstrating the circumstances in which the CDA exception may and may not apply.
A further area of focus is the use of non-legally binding technical measures that allow rights owners to signal their preferences regarding access to and use of their works. Examples include the Robots Exclusion Protocol, metadata-based tools and the EU’s TDM Reservation Protocol. The Consultation Paper takes the view that such measures should remain not legally binding, noting that technological developments and industry practices evolve more rapidly than legislation. However, it seeks feedback on how the development and adoption of these measures can be further supported.
B. Strengthening Copyright Risk Management in AI Deployment and Use
Where an AI-generated output substantially reproduces an existing work, it may infringe copyright. The Consultation Paper recognises that the existing legal principles and framework for assessing copyright infringement apply equally where the allegedly infringing work is generated using AI.
AI developers, deployers, and end users (i.e., the prompters) may each be held liable for copyright infringement depending on the facts of the case and their specific actions. Aside from unauthorised reproduction, liability can arise for unauthorised communication to the public (where the infringing output is also shared or disseminated online) and authorising infringement.
Given the fact-specific nature of copyright infringement claims, uncertainty remains due to the lack of judicial guidance. Accordingly, the Consultation seeks views on how existing copyright principles should apply where AI-generated outputs infringe copyright. In particular, it invites feedback on how responsibility should be apportioned among AI developers, deployers, and end users, with a view to determining measures required to prevent, detect and address infringement risks.
The Consultation Paper also examines the role of technical measures in reducing copyright infringement risks. While noting that some jurisdictions have introduced mandatory safeguards – including the EU General-Purpose AI Code of Practice and China's Cybersecurity Technology Standard for Generative AI Services – Singapore's approach remains focused on encouraging responsible, ethical and trusted AI development, deployment and use. The Consultation therefore seeks feedback on how the development and adoption of effective, proportionate and commercially viable technical measures should be encouraged in Singapore.
C. The Nature of Human Creativity in AI-Assisted Works
As regards copyright subsistence, the Consultation Paper recognises that copyright may subsist in works created with the assistance of generative AI, provided the AI is used as a tool to augment, rather than replace, human creativity. Consistent with Singapore's human-centric approach to copyright, protection is available only where the work is created by a human and constitutes an "authorial creation that is causally connected with the engagement of the human intellect". In practical terms, copyright may subsist in AI-assisted works, but not in works that are entirely AI-generated.
However, this raises a key question: where should the line be drawn between an AI-assisted work that enjoys copyright and an AI-generated work? Put another way, how much human involvement is required for copyright to subsist? The Consultation Paper identifies a range of acts that may demonstrate human creativity, grouped into three stages: (i) pre-generation activities, such as context-setting, grounding and prompting; (ii) post-generation selection and arrangement of AI outputs, including choosing from multiple outputs and combining them into a broader work; and (iii) post-generation editing, involving original human-authored modifications to AI-generated content. Nevertheless, the Consultation seeks feedback on how to distinguish between protectable AI-assisted works and unprotectable AI-generated content, particularly in borderline cases.
Patents
D. Assessing Human Contribution in AI-Assisted Inventions
Patent law in Singapore is likewise human-centric, with only natural persons capable of being named as inventors. Where AI is involved in the inventive process, patent protection remains available, provided that AI functions as an assistive tool and the human researcher identifies the problem and conceives of the invention. By contrast, patent protection is unlikely to be available where an AI system autonomously identifies a problem and generates a novel solution with little or no human involvement.
This naturally raises the question of how to distinguish between an AI-assisted invention and an AI-generated invention. To illustrate this, the Consultation Paper sets out four hypothetical scenarios that address different forms of human involvement, ranging from problem formulation and AI system configuration, to the selection and modification of AI-generated outputs. Through these examples, the Consultation seeks views on how existing inventorship principles should apply, and what level of human contribution should be required for a person to qualify as an inventor.
E. The Effect of AI-Generated Prior Art on Patentability
Finally, the Consultation Paper considers whether and to what extent AI-generated technical disclosures should form part of the prior art landscape. Given AI's ability to generate technical content at scale, such disclosures affect prior art searches and assessments of patentability. Recognising AI-generated disclosures as prior art may narrow the scope of future patent protection and increase the complexity of patent clearance exercises.
Nevertheless, the Consultation Paper also acknowledges practical concerns, including the reliability and technical accuracy of AI-generated content, whether such disclosures are sufficiently enabling, and the increased burden they may place on prior art searches.
Against this backdrop, the Consultation seeks views on whether AI-generated content should qualify as prior art and, if so, whether it should be assessed according to the same standards as human-generated disclosures. It also invites feedback on how the publication date, public accessibility, source and integrity of AI-generated content should be established.
F. Patents and Other AI-Related Issues
Beyond the specific questions outlined above, the Consultation also invites feedback on broader policy issues relating to AI and patents, including the implications of diverging international approaches to the treatment of patentability under each country’s patent law.
Conclusion
The Consultation Paper reflects Singapore's continued preference for adapting existing legal principles to emerging technologies, rather than creating AI-specific intellectual property rights or liabilities. It demonstrates that Singapore's existing intellectual property framework is generally capable of accommodating developments in AI. Key takeaways include:
- AI training will generally continue to benefit from the CDA exception;
- Existing copyright infringement principles are expected to apply to AI-generated outputs; and
- Copyright and patent protection remain contingent on meaningful human contribution.
Nevertheless, significant questions remain at the margins. Consistent with Singapore's broader AI strategy, the Consultation does not seek radical change. Instead, it adopts a measured and stakeholder-driven approach, affirming existing legal principles where possible while inviting feedback on areas of uncertainty. Key areas in which feedback is sought include:
- Support for technical measures governing AI training and access to copyright works.
- Allocation of responsibility for copyright infringement arising from AI-generated outputs.
- The role of technical measures in reducing AI-related copyright infringement risks.
- The distinction between AI-assisted and AI-generated works for copyright purposes.
- The level of human contribution required for inventorship in AI-assisted inventions.
- Whether and how AI-generated disclosures should be treated as prior art.
The Consultation will therefore be an important opportunity for stakeholders to influence how Singapore's intellectual property regime evolves to address the challenges and opportunities presented by AI.
Interested parties may submit their feedback via FormSG or by email to MinLaw at [email protected] by 5pm on 22 October 2026.
We would be happy to assist stakeholders in preparing and submitting their responses to the Consultation Paper.
This article was co-written by Zaine Leong and Ding Ian Toh.
[1] Public Consultation on Artificial Intelligence and Singapore’s Intellectual Property Regime
