GMAsia
    🇸🇬新加坡·AI 新聞·2026年8月27日·來源: Straitstimes

    No need to destroy books in Singapore to legally use them for training AI, say experts

    內容只提供英文版本

    Singapore's legal framework allows AI developers to use lawfully accessed copyrighted works for training AI models, avoiding the controversial practices seen in other countries. This clarity stems from amendments made to Singapore's Copyright Act in 2021. Legal experts, including Tony Yeo of the International Association for the Protection of Intellectual Property's Singapore Group, confirm that this approach provides legal certainty for businesses. This contrasts with situations in the United States, where companies like Anthropic and Amazon have reportedly resorted to physically destroying books to scan their content for AI training, sparking public backlash. The Ministry of Law and Intellectual Property Office of Singapore launched a public consultation on August 26 to further address AI's impact on copyright and patent regimes.

    Nexa 摘要

    Singapore's proactive stance on AI and intellectual property law offers a significant advantage for tech companies operating in Asia. By explicitly permitting the use of legally obtained copyrighted material for AI training, the city-state provides a clear legal framework that reduces uncertainty and encourages innovation. This contrasts sharply with the US, where firms like Anthropic have faced public and legal scrutiny for destructive scanning practices, even as one US court found such use transformative. This legal clarity is crucial for driving AI development, as highlighted by Google's Kent Walker. A regime requiring commercial licenses for every piece of publicly available data would stifle innovation. Singapore, alongside Japan and the European Union, is establishing clear rules for AI training on accessible content, positioning itself as an attractive hub for AI research and development in the region. The ability to use freely viewed online content for AI training, without rights owners being able to exclude their published works, further streamlines development, though AI models must still avoid producing copyrighted infringements in their output.

    原文報道: Straitstimes我們不轉載全文, 閱讀原文 →

    相關文章

    6 則