Govt seeking co-operation with frontier AI firms on transparency and early warning when threats occur: Shanmugam

Singapore and several other countries have engaged frontier artificial intelligence (AI) firms to seek greater transparency and co-operation in addressing future AI-related threats, Coordinating Minister for National Security and Minister for Home Affairs K Shanmugam said on Tuesday (Oct 13).

Senior Minister Shanmugam was at the Singapore International Cyber Week (SICW) at Marina Bay Sands Exhibition and Convention Centre, which has seen attendance from dignitaries of various countries and technology firms.

Speaking with media, SM Shanmugam shared that he has met with major firms including OpenAI, Anthropic, Google and Amazon among others.

He explained that the key ask from Singapore — and many other countries — was the need for transparency immediately, as governments need to know exactly what has happened in real time.

This comes in the wake of a rogue artificial intelligence (AI) agent from ChatGPT maker OpenAI breaching an Australian health-system database earlier this year .

"That was the main ask," he said. "And then of course one of the key purposes of this conference is to try and deepen the partnership between government and industry — and I'm not only talking about Singapore Government."

"We are all now dependent on these major tech companies. They have the technology, they have the platforms, they have the resources, and countries have to work with them."

Shanmugam further stressed that there was a need to "exchange notes" with frontier AI firms, also requesting that they provide an early warning should incidents occur.

Speaking on the SICW, he also said that the meetings he's had and will have over the course of the event present the opportunity to "find common ground" to improve the situation for all countries.

AI cybersecurity tools as a 'powerful weapon system'

Shanmugam also addressed a question on whether cutting-edge AI cybersecurity tools should be made available to governments and the general public.

He shared his perspective that AI has the potential for causing widespread havoc and even physical damage in the hands of "bad people".

"If that is so, then would you automatically make it available broadly, or would you put a framework around it as to who can access it?" He rhetorised.

"It's just like would you give very powerful weapon systems to anyone? There are reasons why these things are controlled, and I think there is a reason for being careful about who gets to access cutting-edge tools and under what conditions."

His perspective was also echoed in a response by the Ministry of Digital Development and Information (MDDI) to Parliamentary questions on Oct 6, evaluating threats that frontier AI models could pose.

"AI becomes more capable of acting on our behalf, we must correspondingly introduce more robust safeguards," MDDI stated then.

"We must be clear about what it is allowed to do, ensure that humans remain in control and accountable where it matters, and continue strengthening our ability to understand emerging risks and respond when problems occur."

Singapore's Cybersecurity Strategy 2026

Shanmugam also launched Singapore's Cybersecurity Strategy (SCS) 2026 at the SICW on Tuesday morning, stating how rapid advances in AI have made cybersecurity immensely challenging.

"There will be more incidents, more attacks," he said. "All of us have to shore up our cyber defences and our resilience to recover when attacks happen."

To do so, Shanmugam highlighted three prongs of the SCS, beginning with the need for Singapore to act earlier.

He emphasised the need for Singapore's defences to be continuous, commenting: "We need to consistently monitor and hunt for threats, identifying weaknesses before adversaries do so, and correcting them quickly."

Next, Singapore also needs to act at scale — to do so, the Government will be enhancing the National Cybersecurity Command Centre (NC3), which was set up in 2024.

NC3 is the Cybersecurity Agency's operations hub for national cyber defence, and aims to provide an understanding of vulnerabilities and threats, detect threat actors and mount a response to these threats.

The enhanced NC3 will strengthen the Cyber Security Agency's ability (CSA) to better anticipate and detect sophisticated cyber threats, strengthen the protection of Critical Information Infrastructure (CII), and coordinate responses to major cyber incidents.

One of the key enhancements is the introduction of a new national tool, known as the Protective Domain Name Service (PDNS), to block access to malware and phishing sites.

This is expected to be rolled out in 2027.

Said Shanmugam: "When someone tries to connect to a domain which is known to be malicious, the PDNS can stop the connection before the threat reaches the user. This will potentially help to reduce exposure to phishing, scams, malware and other cyber threats."

AI to bolster national cyber defence operations

Lastly, Singapore must also act at speed, matching that of the threat.

"If AI allows attackers to identify vulnerabilities and conduct attacks more quickly, then the defenders will have to use AI too," SM Shanmugam explained.

Therefore, the Government will progressively deploy AI across national cyber defence operations, including within NC3 and the Government Cyber Security Operations Centre.

"Singapore will continue to build up our resilience by strengthening our own capabilities and safeguarding our national interests," SM Shanmugam said in closing.

"But resilience does not mean retreating from the world. Far from it. For Singapore, resilience is what gives us the confidence to remain connected to the world."

khooyihang@asiaone.com

新加坡国家安全统筹部长兼内政部长尚穆根周二（10月13日）表示，新加坡和其他几个国家已与前沿人工智能（AI）公司接洽，以寻求在应对未来人工智能相关威胁方面提高透明度和合作。

尚穆根高级部长出席了在滨海湾金沙会展中心举行的新加坡国际网络周（SICW），来自各个国家的政要和科技公司都参加了此次活动。

SM Shanmugam在接受媒体采访时透露，他已经与包括OpenAI、Anthropic、谷歌和亚马逊在内的多家大型公司进行了会面。

他解释说，新加坡以及许多其他国家的主要诉求是立即提高透明度，因为政府需要实时了解究竟发生了什么。

此前，ChatGPT 的制造商 OpenAI 的一个失控人工智能 (AI) 代理于今年早些时候入侵了澳大利亚的一个医疗系统数据库。

“这是我们主要的要求，”他说。“当然，本次会议的关键目的之一是努力深化政府与产业界之间的合作——我指的不仅仅是新加坡政府。”

“我们现在都依赖于这些大型科技公司。它们拥有技术、平台和资源，各国都必须与它们合作。”

尚穆根进一步强调，有必要与前沿人工智能公司“交流意见”，并要求他们在发生事故时提供预警。

在谈到南亚国际合作峰会时，他还表示，他在峰会期间已经举行和将要举行的会议，为“寻找共同点”以改善所有国家的状况提供了机会。

人工智能网络安全工具作为“强大的武器系统”

尚穆根还回答了关于是否应该向政府和公众提供尖端人工智能网络安全工具的问题。

他分享了自己的观点，即人工智能有可能在“坏人”手中造成大范围的破坏，甚至造成人身伤害。

“如果真是这样，那么你们会自动将其广泛开放，还是会设定一个访问框架，规定谁可以访问？”他反问道。

“这就好比你会把威力强大的武器系统交给任何人一样？这些东西受到管制是有原因的，我认为谨慎选择谁能获得尖端工具以及在什么条件下获得这些工具也是有原因的。”

他的观点也得到了数字发展和信息部 (MDDI) 在 10 月 6 日对议会质询的回应，该回应评估了前沿人工智能模型可能构成的威胁。

“随着人工智能越来越有能力代表我们行事，我们必须相应地引入更强有力的保障措施，”MDDI当时表示。

“我们必须明确哪些行为是被允许的，确保在关键时刻人类仍然拥有控制权和责任感，并不断加强我们理解新出现的风险并在问题发生时做出应对的能力。”

新加坡2026年网络安全战略

尚穆根周二上午在新加坡网络安全周上发布了新加坡网络安全战略（SCS）2026，并指出人工智能的快速发展给网络安全带来了巨大的挑战。

他说：“还会发生更多事件，更多攻击。我们所有人都必须加强网络防御，提高应对攻击的能力。”

为此，尚穆根强调了南海问题的三个方面，首先是新加坡需要更早采取行动。

他强调新加坡的防御需要持续不断，并评论道：“我们需要持续监控和搜寻威胁，在对手发现弱点之前就发现弱点，并迅速纠正它们。”

接下来，新加坡还需要采取大规模行动——为此，政府将加强于 2024 年成立的国家网络安全指挥中心 (NC3)。

NC3 是网络安全局的国家网络防御运营中心，旨在帮助人们了解漏洞和威胁，检测威胁行为者并对这些威胁做出响应。

增强型 NC3 将增强网络安全局 (CSA) 更好地预测和检测复杂网络威胁的能力，加强对关键信息基础设施 (CII) 的保护，并协调对重大网络事件的响应。

其中一项关键改进是引入了一种名为“保护性域名服务 (PDNS)”的新型国家工具，以阻止对恶意软件和网络钓鱼网站的访问。

预计将于2027年推出。

Shanmugam表示：“当有人试图连接到已知恶意域名时，PDNS可以在威胁到达用户之前阻止连接。这有助于减少用户遭受网络钓鱼、诈骗、恶意软件和其他网络威胁的风险。”

人工智能将加强国家网络防御行动

最后，新加坡也必须迅速采取行动，与威胁的速度相匹配。

“如果人工智能能够让攻击者更快地发现漏洞并进行攻击，那么防御者也必须使用人工智能，”SM Shanmugam解释道。

因此，政府将逐步在国家网络防御行动中部署人工智能，包括在国家网络安全中心和政府网络安全行动中心内。

“新加坡将继续通过增强自身能力和维护国家利益来增强我们的韧性，”尚穆根总理在总结发言中说道。

“但韧性并不意味着与世隔绝。恰恰相反。对新加坡而言，韧性赋予我们有信心与世界保持联系。”

khooyihang@asiaone.com