Singapore is asking its citizens and industry stakeholders to weigh in on one of the most contentious questions in AI policy today: should technology companies be allowed to use copyrighted books, music, news articles, and images to train artificial intelligence models without paying creators or getting their permission? The government launched a formal public consultation on August 26 to gather feedback on how copyright and patent laws should evolve in the age of AI.

The timing reflects a global surge in copyright lawsuits against AI companies. Rights holders, including authors, musicians, and publishers, have filed cases alleging that tech firms have infringed their copyrights by using their creative works to train AI models without consent or compensation. Singapore's Ministry of Law and Intellectual Property Office of Singapore (IPOS) want to understand how their country should navigate this legal minefield before disputes escalate domestically.

What Do Courts Around the World Say About AI and Copyright?

The answer depends heavily on the specific facts of each case. In the United States, courts have reached strikingly different conclusions about the same company. AI firm Anthropic, which developed the Claude AI chatbot, faced a significant ruling: a judge ordered the company to pay US$1.5 billion (approximately S$1.9 billion) to affected authors for infringing their copyrights when it used millions of pirated e-books to train its AI without consent.

However, the same judge found that Anthropic did not infringe copyright when it trained its AI by scanning pages from old physical books the company had purchased, then destroyed and recycled. The court determined this destructive scanning was "transformative" and qualified as fair use, even though the books included rare and out-of-print titles. This contradiction illustrates why Singapore's government believes clarity is urgently needed.

Why Is Singapore Taking Action Now?

Minister for Law Edwin Tong explained that the issue will only become more pressing as AI becomes more powerful and widely used as a creative tool. AI models require increasingly large amounts of training data to function effectively, which raises fundamental questions about authenticity and creativity.

"Courts and IP offices around the world broadly agree that an author must be a human being. But where do we draw the line? What kinds of acts or decisions made by human prompters demonstrate sufficient creativity?" said Edwin Tong.

Edwin Tong, Minister for Law and Second Minister for Home Affairs, Singapore

Tong emphasized that while the government wants to promote innovation and avoid stifling technological progress, the rights of creators must be carefully balanced. He noted that it would not be sustainable in the long term if creative works could be used at scale by AI systems without a framework that respects their creation and compensates their creators.

What Specific Issues Is Singapore Consulting On?

The public consultation addresses several interconnected questions about how copyright law should adapt to AI:

• Training Data Permissions: Whether copyrighted works can legally be used to train AI models without explicit permission from or payment to rights holders.
• Creator Protections: How copyright protections should be enforced when creators do not want their work used for AI training, including the role of technical measures and who bears the cost of implementing them.
• Invention and Patents: How existing principles on inventions should apply when AI is used in the invention process, raising questions about ownership and attribution.
• Accountability Mechanisms: Ways to foster greater accountability and certainty regarding copyrights when training AI, ensuring both transparency and fair compensation.

How Can Creators Protect Their Work From AI Training?

Tong acknowledged that the rise in litigation suggests practical questions remain unanswered about how rights holders should protect their work if they do not want it used for AI training. One potential solution involves deploying more technical measures, such as digital rights management or metadata that signals whether a work should be excluded from AI training datasets.

However, this approach raises equity concerns. If technical protection becomes the standard, the government must consider how to support less well-resourced rights holders, such as independent authors and small publishers, who may lack the resources to implement such protections. Tong stated that the government "will carefully study the global landscape, and consider an approach which gives us the right balance".

Tong

The public consultation also seeks feedback on how to value intangible assets, including copyrights and patents. The government has already commissioned guidelines on this topic, developed by the Institute of Valuers and Appraisers Singapore with support from the Accounting and Corporate Regulatory Authority, IPOS, and industry partners. These guidelines aim to build greater consistency and confidence in how intellectual property assets are valued.

What Happens Next?

Stakeholders, particularly those in creative industries, have until 5 p.m. on October 22 to submit feedback to the government. Details on the public consultation can be found at go.gov.sg/ai-ip2026. Singapore's approach will likely influence how other countries in the Asia-Pacific region address similar questions, as the region becomes an increasingly important hub for AI development and creative industries.

The consultation reflects a broader global recognition that copyright law, written decades before AI existed, may not adequately address the challenges posed by machine learning. Singapore's willingness to engage the public in this debate signals that policymakers recognize this is not a question that should be decided by courts and lawyers alone, but one that requires input from creators, technologists, and society at large.