Singapore launches consultation on AI training, copyright, and patents

Singapore has launched a major public consultation examining how artificial intelligence should interact with the country’s intellectual property regime, including the use of copyrighted works to train AI systems and the treatment of inventions created with AI assistance.

The Ministry of Law (MinLaw) and Intellectual Property Office of Singapore (IPOS) launched the consultation on August 26, 2026, with responses invited until 5:00 p.m. on October 22, 2026.

The consultation comes as governments and courts around the world grapple with the rapidly developing legal questions surrounding AI training, copyrighted works and AI-assisted invention.

Singapore said it is seeking views on whether its IP laws and frameworks remain “fit for purpose” as AI changes how people create, invent, research, and conduct business.

The consultation covers both copyright and patents, with the government seeking feedback on proposed legal refinements, practical information resources, and non-binding technical measures intended to increase certainty and accountability.

Can copyrighted works be used to train AI?

One of the consultation’s central issues is whether Singapore’s existing copyright framework provides sufficient certainty around AI training.

Singapore already has a specific exception for computational data analysis. Under the Copyright Act, lawfully accessed copyright works can be used for computational data analysis, including text and data mining and training machine-learning systems, without requiring permission from each individual copyright owner.

The new consultation, therefore, does not start with a blank slate.

Instead, MinLaw and IPOS are asking whether the scope and application of that exception are sufficiently clear, how the requirement for lawful access and safeguards for rights owners should operate in practice, and how non-binding technical measures could be supported in an industry-led and practical way.

That puts Singapore at the center of one of the most contentious questions in global IP law:

When can an AI developer use copyrighted material to train an AI model — and what rights should creators have to control or protect their works?

The question has become increasingly urgent as copyright owners pursue litigation against AI companies in multiple jurisdictions.

In the US, for example, Anthropic was ordered to pay $1.5 billion to authors over its use of millions of pirated e-books to train its AI systems. The same case also produced a different finding concerning books Anthropic had lawfully purchased and subsequently scanned.

The Singapore government noted that courts overseas have reached different conclusions on whether copyright is infringed when protected works are used to train AI without permission, depending on the circumstances of individual cases.

Accountability for AI-generated output

The consultation also moves beyond the question of training data.

MinLaw and IPOS are seeking views on copyright risk management in AI deployment and use, including how existing legal principles should apply where AI-generated output infringes copyright.

The government is also examining how responsibility should be assessed among AI developers, deployers, and end users, together with the potential role of technical measures designed to reduce the risk of infringing output.

Singapore is seeking views on whether such measures can be implemented in a way that is both proportionate and commercially viable.

That could become increasingly important as businesses incorporate generative AI into everyday creative and commercial workflows.

Where does human creativity end?

A third copyright issue concerns the nature of human creativity in AI-assisted works.

The consultation asks how human contribution should be understood, recognized and evidenced when generative AI is used as a tool in creative and business workflows.

The issue is likely to become increasingly important for companies using AI to generate images, advertising material, written content and other commercial assets.

Singapore Law Minister Edwin Tong highlighted the issue at the Global Forum on Intellectual Property during Singapore IP Week.

Tong said that courts and IP offices around the world broadly agree that an author must be a human being, but questioned where the line should be drawn when humans use AI as part of the creative process.

The question, he suggested, is what kinds of decisions or acts by a human using AI demonstrate sufficient creativity.

AI and the future of patent inventorship

The consultation also addresses patents.

MinLaw and IPOS are asking how existing inventorship principles should apply across different forms of human-AI interaction during the inventive process.

The consultation specifically identifies activities including problem formulation, selecting from AI-generated outputs and modifying AI-generated technical solutions.

The government is also examining the potential impact of the large-scale publication of AI-generated technical disclosures.

According to the consultation, such material could affect the prior-art landscape, including patent searches, examination, and patentability assessments. It could also have implications for incentives to innovate as increasing amounts of AI-generated technical material become publicly available.

The issue adds another dimension to the international debate over whether existing patent law is capable of dealing with increasingly sophisticated AI-assisted invention.

Singapore joins global AI-IP race

Singapore’s consultation places it alongside other major jurisdictions attempting to establish workable rules for AI and intellectual property.

The UK has already undertaken a major consultation on copyright and AI, examining issues including the use of copyright works for AI training, rights-holder control, and licensing. The UK government published its consultation response and proposals in 2026.

In the EU, the AI Act requires providers of general-purpose AI models to put in place a copyright policy and publish a summary of the content used to train their models.

The US, meanwhile, continues to address many AI-training disputes through litigation and the application of existing copyright doctrines, producing case-specific decisions rather than a single comprehensive AI-training regime.

Singapore’s approach is notable because it is considering how to refine an existing copyright framework that already permits certain forms of AI training.

The consultation could therefore help determine whether the country’s computational data analysis exception remains appropriate as AI systems become more powerful and the commercial value of training data continues to grow.

Implications for IP owners

Although the consultation is specifically focused on copyright and patents, its implications extend across the wider IP landscape.

For brand owners, AI is increasingly being used to generate advertising, images, product descriptions and other brand-related material. Questions surrounding training data, human creativity, and responsibility for AI-generated content could therefore increasingly intersect with broader IP management and enforcement strategies.

For patent owners and innovators, the consultation raises potentially significant questions about who qualifies as an inventor when AI plays a substantial role in the inventive process and how AI-generated technical material should be treated when assessing prior art.

Singapore’s move is therefore another important indication that governments are beginning to confront the practical IP consequences of generative AI rather than relying solely on existing legal principles and litigation to resolve them.

The consultation period runs until 5:00 p.m. on October 22, 2026. Responses can be submitted to MinLaw through separate forms for copyright and patents or by email.

Singapore Ministry of Law: Public Consultation on Artificial Intelligence and Singapore’s Intellectual Property Regime

← Back to News