Skip to main content

Singapore sets up expert task force to counter AI-driven cyber threats

A report is expected in the first half of 2027, while new initiatives cover testing autonomous AI systems and accrediting independent testers.

By Arif Rahman

Illustration of connected computer systems and digital security controls
Photo: CNA

Singapore has established a group comprising more than 20 industry leaders and experts to improve cyber defences in its infocomm and media sector as AI enables and accelerates attacks. The Infocomm Media Development Authority (IMDA) announced the initiative on Oct 14, alongside a testing kit for agentic AI systems.

Named the Frontier AI Resilience and Security Taskforce (FAIRS), the group includes specialists from telecommunications, media, cybersecurity and AI. Its priorities are faster threat responses, stronger protection for critical systems and shared intelligence to detect new attack methods earlier. A report offering practical guidance and recommended practices is expected during the first half of 2027.

IMDA chief executive Ng Cher Pong said testing AI applications individually could not address every risk. The task force will examine network and system segmentation to limit disruption from vulnerabilities, alongside improvements to network visibility and response capabilities.

IMDA's new starter kit addresses the safety and reliability of AI agents, which can use tools, access sensitive information and external systems, and act for users. It recommends assessing an agent's reasoning and actions throughout a task, rather than judging only the final result.

The kit outlines four stages: identifying risks, creating realistic tests, examining logs, and ongoing monitoring and retesting. It incorporates input from more than 20 industry deployers and testers, plus work through the Global AI Assurance Sandbox and Singapore AI Safety Institute. IMDA will revise it as testing approaches develop.

Separately, the AI Verify Foundation announced its first five accredited firms under the AI Tester Accreditation Programme: AIDX Tech, Knovel Engineering, PwC Singapore, Resaro and Vulcan. Accreditation runs for one year and recognises demonstrated testing competencies, rather than an ability to assess every risk. Assessments cover technical skills, past work, delivery capacity and financial sustainability.

Keep reading: AI

All AI stories

Latest in Technology

All latest