口頭答覆 · 2026-02-12 · 屆國會 15
新加坡AI經濟與資料主權
議員質詢政府在推動AI經濟及全球領導地位上的預算安排,探討是否設立政府AI公司以保障資料主權及公共利益。政府回應強調國家AI戰略2.0,注重資料安全、技術效能及國際合作。核心爭議在於如何防止全球科技巨頭主導本地AI市場,避免資料流失及技術依賴,及對外資使用公共資料的監管和收益分配問題。
關鍵要點
- • 政府AI戰略投入
- • 資料主權與安全
- • 防止技術依賴
推動AI發展兼顧資料安全與國際合作
擔憂外企主導資料與技術依賴
加強資料主權與智慧財產權保護
“We have robust data protection framework to facilitate the secure processing of data.”
參與人員 (3)
- Gerald Giam Yean Song
- Senior Minister of State for Digital Development and Information
- Tan Kiat How
完整譯文(中文)
Hansard 原始記錄 · 2026-05-02
11號議員嚴彥松先生詢問數字發展與信息部長:(a) 政府在全球增長領域領導力和人工智慧賦能經濟戰略上的預計支出是多少;(b) 政府是否會通過政府擁有的人工智慧公司資助具有突破性增長潛力的雄心勃勃的探索性專案;(c) 政府是否認為此類實體在追求資料主權和公共利益至關重要的專案時,比外國實體更有效。
數字發展與資訊高階國務部長陳杰豪先生(代表數字發展與信息部長)答覆:根據新加坡國家人工智慧(AI)戰略2.0,我們致力於利用人工智慧造福公眾,改善新加坡人的生活。該願景通過在公共和私營部門的投資支援,建設研究、政府和產業的人工智慧能力,並將人工智慧應用於解決現實世界問題來實現。
在發展這些能力時,我們考慮符合新加坡需求的因素,同時兼顧技術的效能、安全性、韌性和成本效益。我們充分重視資料安全和公共利益。我們利用本地能力,並與國際實體合作,包括能夠傳授知識、培養技能和創造本地就業的領先公司。
議員可參考即將釋出的2026年預算宣告和撥款委員會辯論,瞭解新加坡如何發展人工智慧賦能經濟的更多細節。
議長:嚴議員。
嚴彥松議員(阿裕尼):感謝高階國務部長的答覆。部長先生,政府是否擔心我們的人工智慧領域將被全球科技巨頭主導?政府將如何防止所謂的“資料流失”,即這些公司利用我們的公共資料或公共資助構建更優的專有模型,最終向新加坡使用者收費訂閱,或使本地企業陷入長期技術依賴的境地?
其次,高階國務部長是否會考慮一項政策,要求任何使用新加坡公共資料集進行人工智慧訓練的外國公司必須承諾實施結構化的本地知識轉移計劃,並與新加坡企業共享部分由此產生的智慧財產權?
最後,鑑於公共資料集是國家資源,政府是否會考慮向商業人工智慧開發者,尤其是外國企業,收取使用這些資料集訓練模型的費用?這筆收入將如何核算和公開,以確保其用於公共利益的實現?
陳杰豪:部長先生,無論公司是在本地註冊還是在新加坡境外註冊,我們都有明確且健全的資料保護框架,以促進資料的安全處理。
對於私營部門,所有組織必須遵守《個人資料保護法》下的義務。這些義務包括保護義務,即保護其持有的個人資料免遭未經授權的訪問,以及轉移限制義務,該義務適用於資料轉移至其他國家的情況,無論這些資料是用於為新加坡客戶處理,還是用於其他目的,包括訓練其人工智慧模型的資料。
這些組織還必須遵守適用於其行業的其他資料法規。正如議員所提及的,政府作為一個行業也適用這些法規,此外還有金融、電信、醫療保健、物流等行業,這些行業可能有其特定的要求。
本議院最近就《健康資訊法案》二讀以及《公共部門治理法》進行了辯論,這些法案施加了類似的要求和保障,以確保高標準的資料安全。此類資料(議員提及的公共部門資料)只有在有合法目的支援公共利益,並且具備部長授權和資料保護及安全的合同協議時,才能共享。
關於議員提出的商業模式具體細節,這些必須根據具體用例和相關負責機構的需求來確定。
英文原文
SPRS Hansard · Fetched: 2026-05-02
11 Mr Gerald Giam Yean Song asked the Minister for Digital Development and Information (a) what is the projected expenditure on the Government’s strategies for global leadership in growth areas and an AI-empowered economy; (b) whether the Government will fund ambitious exploratory projects with breakout growth potential through a Government-owned AI corporation; and (c) whether the Government considers such entities more effective than foreign entities at pursuing projects where data sovereignty and public interest are paramount.
The Senior Minister of State for Digital Development and Information (Mr Tan Kiat How) (for the Minister for Digital Development and Information) : Under Singapore's National Artificial Intelligence (AI) Strategy 2.0, we seek to harness AI for the public good and to improve the lives of Singaporeans. This vision is supported by making investments across both the public and private sectors, building up AI capabilities in research, government and industry, and applying AI to solve real-world problems.
In developing these capabilities, we consider what serves Singapore's needs, alongside other factors, such as the performance, security, resilience and cost-effectiveness of the technology. There is due regard for data security and the public interest. We leverage local capabilities and partner with international entities, including leading companies who can impart knowledge, build skills and create local jobs.
The Member may refer to the upcoming 2026 Budget Statement and the Committee of Supply debates for more details on how Singapore will develop an AI-empowered economy.
Mr Speaker : Mr Giam.
Mr Gerald Giam Yean Song (Aljunied) : I thank the Senior Minister of State for the reply. Sir, does the Government have any concerns that our AI landscape will be dominated by global technology giants? How will the Ministry prevent what is called a "data drain" where these companies use our public data or public grants to build superior proprietary models, and that they eventually charge Singaporean subscriptions for or risk forcing local firms into a position of enduring technical dependency?
Secondly, will the Senior Minister of State consider a policy where any foreign firm using Singapore's public data sets for AI training must commit to a structured local knowledge transfer programme and share a portion of the resulting intellectual property with Singaporean firms?
And lastly, given that public data sets are a national resource, will the Government consider charging commercial AI developers, particularly foreign-based ones, a fee for training their models on these sets and how will this revenue be accounted for and publicised to ensure it is used for the delivery of public goods?
Mr Tan Kiat How : Sir, regardless of whether the companies are domiciled here or are domiciled outside Singapore, I think it is very clear that we have robust data protection framework to facilitate the secure processing of data.
For the private sector, all organisations must comply with their obligations under the Personal Data Protection Act. Examples of these obligations include the Protection Obligation to protect personal data in their possession from unauthorised access and the Transfer Limitation Obligation, which applies when data is transferred to another country, regardless of whether they use the data for processing it for Singapore clients, or for something else, including training data for their AI models.
These organisations must also comply with any additional data regulations which apply to their sectors, and as the Member has alluded to, that includes the Government as a sector, but there are also financial sectors, telecommunications, healthcare, logistics and so forth. These sectors may have its own specific set of requirements.
In this House, we just had a debate on the Second Reading of the Health Information Bill as well as the Public Sector Governance Act, which impose similar requirements and safeguards to ensure high standards of data security. Such data can only be shared, and that is for the public sector data that the Member has mentioned if there is a legitimate purpose to support the public interest, and the Ministerial authorisation and contractual agreements for data protection and data security are in place.
His other questions around the specifics of the commercial model, the business model, these have to be use case specific and driven by the use case problems and the relevant agencies looking after it.