The\u00a0Singapore\u00a0public sector\u00a0has been laser-focused in embracing\u00a0artificial intelligence\u00a0(AI),\u00a0with\u00a080 per cent of the 150,000 public officers<\/a>\u00a0already using AI tools in their workplace, one of the world\u2019s highest adoption rates.\u00a0<\/p>\n \u00a0<\/p>\n However, the biggest potential road\u00a0bump\u00a0to this blue-sky\u00a0vision of\u00a0an\u00a0AI-enabled\u00a0smart\u00a0nation\u00a0was\u00a0Mythos<\/em>,\u00a0the Greek word for myth or mythology.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n Claude\u00a0Mythos\u00a0was\u00a0Anthropic\u2019s\u00a0latest\u00a0AI model, which has shown an\u00a0unprecedented ability to both detect and exploit vulnerabilities across critical software systems<\/a>.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n More on Mythos later.\u00a0<\/p>\n \u00a0<\/p>\n Coming back to Singapore\u2019s AI ambition, the Singapore Prime Minister and Minister for Finance, Lawrence Wong, in 2024\u00a0during the launch of the\u00a0country\u2019s\u00a0updated\u00a0Smart Nation 2.0 policy<\/a>, articulated the country’s ambition.\u00a0<\/p>\n \u00a0<\/p>\n Known for being acutely aware of both its strengths and weaknesses, Singapore has\u00a0identified\u00a0the need for a pool of talented\u00a0manpower\u00a0well conversant with AI\u00a0in order to\u00a0realise its Smart Nation ambition.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n As a result, a major focus has been steps to\u00a0augment\u00a0AI talent in the nation.\u00a0<\/p>\n \u00a0<\/p>\n The\u00a0government\u2019s latest announcement that\u00a0it plans\u00a0to\u00a0upskill 40,000 tech professionals<\/a>\u00a0by 2029\u00a0was the latest in a series of\u00a0policy moves to\u00a0dramatically increase the\u00a0number\u00a0of\u00a0AI enabled workers.\u00a0<\/p>\n \u00a0<\/p>\n This move complemented\u00a0the\u00a0earlier announced\u00a0initiative to train 100,000 non-technology professionals\u00a0to be AI-bilingual by 2029.\u00a0<\/p>\n \u00a0<\/p>\n The excitement about the possibilities\u00a0that open with the widespread use of\u00a0AI\u00a0has been\u00a0quite palpable,\u00a0especially in the public sector.<\/a>\u00a0\u00a0<\/p>\n \u00a0<\/p>\n Ironically,\u00a0the companies that have been\u00a0at the forefront of\u00a0the large language models (LLMs),\u00a0which\u00a0are the brains behind AI\u00a0systems,\u00a0have\u00a0sounded\u00a0the alarm\u00a0in this path to AI\u00a0Nirvana<\/em>.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n They appear fearful of their own creation!\u00a0<\/p>\n \u00a0<\/p>\n A recent report by\u00a0Google Threat Intelligence Group (GTIG)\u00a0said\u00a0that\u00a0AI-powered hacking<\/a>\u00a0has gone from a nascent problem to an\u00a0\u201cindustrial-scale threat\u201d.\u00a0<\/p>\n \u00a0<\/p>\n Echoing the\u00a0Google\u00a0warning,\u00a0Palo Alto Networks\u2019 Chief Product and Technology Officer\u00a0Lee Klarich\u00a0has\u00a0said<\/a>\u00a0organisations\u00a0have a narrow three-to-five-month window \u201cto outpace the adversary before AI-driven exploits start to become the new norm.\u201d\u00a0<\/p>\n \u00a0<\/p>\n Most of these warnings can be traced to\u00a0an\u00a0unusual\u00a0announcement\u00a0by\u00a0Anthropic\u00a0in April.\u00a0<\/p>\n \u00a0<\/p>\n The company\u00a0took the unprecedented step of issuing a\u00a0public\u00a0warning about\u00a0Claude Mythos.\u00a0<\/p>\n \u00a0<\/p>\n Anthropic\u00a0declared the\u00a0model was \u201ctoo powerful for public release\u00a0due to its unprecedented ability to identify software vulnerabilities\u201d,\u00a0claiming<\/a>\u00a0that\u00a0the\u00a0model was able find vulnerabilities in systems that had been overlooked by human reviews and security tests.\u00a0<\/p>\n \u00a0<\/p>\n In\u00a0a\u00a0blog post<\/a>, Anthropic\u00a0said\u00a0the\u00a0model\u00a0was\u00a0able\u00a0find vulnerabilities in systems that had\u00a0been overlooked for more than 20 years within a few minutes.\u00a0<\/p>\n \u00a0<\/p>\n The company\u00a0was scared enough to\u00a0restrict access to the model to\u00a0just\u00a040 tech and security companies under\u00a0Project\u00a0Glasswing<\/a>.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n The ostensible reason\u00a0was\u00a0to\u00a0give\u00a0these companies\u00a0time\u00a0to use Mythos to check on security vulnerabilities in existing networks and, in a\u00a0best-case\u00a0scenario, fix them\u00a0as soon as possible\u00a0before a public release.\u00a0<\/p>\n \u00a0<\/p>\n Unfortunately, software\u00a0was\u00a0like a genie in a bottle; once it came out, there was no putting it back in\u00a0and they have\u00a0a notorious habit of leaking out into the wild.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n Experts\u00a0were\u00a0not talking about\u00a0whether cyber criminals will get their hands on a version of Mythos, they\u00a0were\u00a0focussing on\u00a0how soon this will happen.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n It was also\u00a0only a matter of time before other AI models\u00a0would\u00a0catch\u00a0up with Mythos.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n This explained\u00a0the\u00a0warnings issued\u00a0by Google and Palo Alto Networks, among others.\u00a0<\/p>\n \u00a0<\/p>\n This brings us back to\u00a0Singapore\u2019s rapid embrace of AI, especially in the public sector.\u00a0<\/p>\n \u00a0<\/p>\n How prepared\u00a0was\u00a0the Republic to face\u00a0this new wave of AI-driven cyberattacks, often\u00a0orchestrated\u00a0by\u00a0state actors<\/a>?\u00a0\u00a0<\/p>\n \u00a0<\/p>\n The answer would\u00a0be found\u00a0only\u00a0once an actual attack occurred,\u00a0but\u00a0the indications were that the\u00a0government\u00a0at the leadership and policy level\u00a0was\u00a0well\u00a0aware\u00a0of the threat posed by Mythos and its analogues.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n Earlier this month,\u00a0speaking in Parliament<\/a>, Singapore\u00a0Senior Minister of State for Digital Development and Information,\u00a0Tan Kiat How,\u00a0acknowledged\u00a0the dangers posed by Mythos\u00a0in detecting\u00a0previously undiscovered\u00a0software vulnerabilities.\u00a0<\/p>\n \u00a0<\/p>\n Tan noted that while the\u00a0government\u00a0was\u00a0yet seen fully autonomous AI agents conducting end-to-end campaigns,\u00a0\u201cit\u00a0was\u00a0a matter of time\u201d.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n To allay fears, he emphasised that\u00a0what\u00a0Mythos represented should be\u00a0seen as\u00a0a “continuum” of developments\u00a0rather than a sudden, singular event.\u00a0<\/p>\n \u00a0<\/p>\n On May 9,\u00a0speaking at a public event<\/a>,\u00a0Coordinating Minister for National Security K. Shanmugam, noted that cyber\u00a0attackers\u00a0were\u00a0targeting Singapore by\u00a0using AI to make\u00a0attacks cheaper and faster.<\/p>\n \u00a0<\/p>\n He\u00a0added\u00a0that this was\u00a0a serious\u00a0issue,\u00a0and operators of critical information infrastructure (CIIs)\u00a0urgently needed to improve their\u00a0defence\u00a0preparedness.\u00a0<\/p>\n \u00a0<\/p>\n On the\u00a0operations\u00a0front, a slew of rules, regulations and advisories have been issued\u00a0to protect\u00a0systems that used AI, even before Mythos came on the scene.\u00a0<\/p>\n \u00a0<\/p>\n Early this year, the\u00a0Infocomm\u00a0Media Development Authority (IMDA)\u00a0of Singapore\u00a0launched the\u00a0Model AI Governance Framework for Agentic AI (MGF)<\/a>.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n This\u00a0was\u00a0the world\u2019s first governance framework specifically designed for AI agents capable of autonomous planning,\u00a0reasoning\u00a0and action.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n The policy focussed\u00a0on ensuring\u00a0that the risks were assessed upfront and there would be human-in-the-loop\u00a0while implementing technical controls and processes. It also enabled end-user responsibility.\u00a0<\/p>\n \u00a0<\/p>\n The MGF has built\u00a0on\u00a0earlier governance instruments for traditional\u00a0AI\u00a0and generative AI\u00a0(GenAI) policies\u00a0to address novel risks arising\u00a0from the\u00a0autonomous agentic AI\u00a0bots.\u00a0<\/p>\n \u00a0<\/p>\n Apart from this, the Cyber Security Agency of Singapore (CSA)\u00a0has released<\/a>\u00a0discussion papers and addendums specifically focused on securing agentic AI systems,\u00a0providing\u00a0practical guidance on mapping workflows and applying controls across the entire AI lifecycle.\u00a0<\/p>\n \u00a0<\/p>\n The Singapore government\u00a0has also been\u00a0investing in AI-powered tools for<\/p>\n active vulnerability and patch testing.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n This included\u00a0in-house capability building and partnerships to adapt global tools for detecting, triaging, and responding to threats at “machine speed”.\u00a0<\/p>\n \u00a0<\/p>\n All these developments\u00a0showed\u00a0that the government was\u00a0aware of the risks of AI.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n Being aware of the risks is\u00a0half the job done. Being able to prevent attacks was the other\u00a0and more important\u00a0half.\u00a0<\/p>\n \u00a0<\/p>\n This is where things start to get uncertain. In\u00a0cybersecurity a network\u2019s strength was\u00a0decided\u00a0by its\u00a0most vulnerable point.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n And that vulnerable point was the end-user.\u00a0Most breaches\u00a0in cybersecurity\u00a0occur due to\u00a0actions taken by humans<\/a>.\u00a0<\/p>\n \u00a0<\/p>\n This\u00a0was\u00a0important because as the country rushed towards\u00a0training\u00a0more professionals in the use of AI, the focus\u00a0should\u00a0also be on\u00a0making them aware of the\u00a0risks.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n The glamour of AI and autonomous bots should not hide the potential dangers.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n The ultimate success of Singapore\u2019s pivot into an\u00a0AI-first\u00a0nation would\u00a0depend not only\u00a0just\u00a0on\u00a0developing\u00a0an\u00a0AI savvy\u00a0workforce; it would depend on\u00a0a\u00a0workforce\u00a0aware of\u00a0both the\u00a0advantages and dangers of AI<\/em>.\u00a0<\/p>\n \u00a0<\/p>\n This required mandatory training on cyber hygiene and\u00a0on how to defend in case of an attack.\u00a0Cybersecurity was always important but in the new age of AI it has become a major imperative.\u00a0<\/p>\n \u00a0<\/p>\n The country\u2019s\u00a0Smart Nation 2.0<\/a>\u00a0journey\u00a0hinged\u00a0on developing this awareness.\u00a0<\/p>\n \u00a0<\/p>\nDevelopers sound the alarm\u00a0<\/h3>\n
How prepared is Singapore?\u00a0<\/h3>\n
New rules and regulations\u00a0<\/h3>\n
Network strong as weakest link\u00a0<\/h3>\n
![\"\"](\"https:\/\/gov-web-sing.s3.ap-southeast-1.amazonaws.com\/uploads\/2026\/2\/image-1772011726968.jpg\")
<\/a>