SINGAPORE: Singapore has formed a task force of more than 20 experts and industry leaders to strengthen the infocomm and media sector's defences against AI-enabled and AI-accelerated cyber threats.
The Frontier AI Resilience and Security Taskforce (FAIRS) brings together leaders and experts from the telecommunications, media, cybersecurity and artificial intelligence (AI) sectors, the Infocomm Media Development Authority (IMDA) said on Wednesday (Oct 14).
Its work will focus on three areas:
- Improving the speed of responses to AI-accelerated threats.
- Strengthening critical systems as AI becomes better at uncovering vulnerabilities.
- Pooling intelligence across the sector to identify emerging attack techniques earlier.
The task force is expected to publish a report with practical guidelines and best practices in the first half of 2027.
"Frontier AI is changing the rules of the game and we must adapt accordingly," IMDA chief executive Ng Cher Pong said at the AI Assurance Exchange: Agentic Edition, an industry event.
He added that some risks could not be managed by testing individual AI applications alone, particularly as frontier AI changes the cybersecurity threat landscape.
For instance, the task force will examine how networks and systems can be strengthened and segmented to prevent vulnerabilities from leading to widespread disruption, as well as how network visibility and response capabilities can be improved to keep pace with faster-moving threats.
The announcement comes a day after Singapore launched a new national cybersecurity strategy aimed at enabling earlier and faster responses to AI-enabled threats, including by progressively deploying AI across national cyber defence operations.
TESTING AGENTIC AI SYSTEMS
IMDA also launched the first version of a starter kit aimed at helping organisations test agentic AI systems for safety and reliability.
Unlike conventional generative AI systems, AI agents can act on behalf of users, use tools, access external systems and sensitive information, and make changes to their environment.
"When we give AI more agency, we need greater assurance over how that agency is exercised," Mr Ng said.
He added that testing such systems needed to look beyond whether an AI agent ultimately arrived at the correct outcome.
Organisations should also examine what tools an agent used, what information it accessed, the decisions it made along the way, whether it respected the permissions and boundaries it was given and whether safeguards worked as intended.
The starter kit sets out a four-step testing process: identifying risks such as erroneous or unauthorised actions, designing realistic tests, analysing logs and continuously monitoring and re-testing systems.
It also calls for testing an agent's entire trajectory, including its reasoning and actions, rather than looking only at its final output.
The starter kit draws on consultations with more than 20 industry deployers and testers, as well as testing work through the Global AI Assurance Sandbox and Singapore AI Safety Institute. It will be updated as testing methods and best practices evolve.
"As AI evolves rapidly, testing methods have to catch up too. The Starter Kit for Testing Agentic AI Systems for Safety and Reliability is therefore a living document, which we will continue to update and refine, and we welcome your feedback along the way," Mr Ng said.
The launch comes a week after Senior Minister of State for Digital Development and Information Tan Kiat How said the government was running trials and experiments to understand what it takes to implement agentic AI systems responsibly.
He had said the government was also reviewing how to strengthen existing frameworks and systems to account for the increasing autonomy of AI agents, while monitoring developments overseas.
The launch also follows high-profile incidents earlier this year where AI agents went "rogue", including swarms of hundreds of AI agents from industry giant OpenAI that hacked AI infrastructure firm Hugging Face. The agents acted outside their brief, coordinated with one another and tried to cover their tracks.
ACCREDITED AI TESTERS
Separately, the AI Verify Foundation named the first five firms accredited under its AI Tester Accreditation Programme (AI TAP).
They are: AIDX Tech, Knovel Engineering, PwC Singapore, Resaro and Vulcan.
The programme was announced in May and aims to help businesses identify third-party firms capable of testing the safety and reliability of generative AI applications.
Companies seeking accreditation are assessed on areas including their technical competency, track record, delivery capability and financial sustainability.
Accreditation lasts for one year, allowing standards to be reviewed as AI technology evolves.
"Accreditation is not intended to suggest that one firm can test everything," Mr Ng said.
Instead, firms are recognised for the specific testing competencies they have demonstrated, giving enterprises greater clarity over which testers they can engage for particular risks, he added.
The three initiatives address different parts of the AI assurance landscape:
- Testing individual agentic systems.
- Establishing a pool of accredited independent testers.
- Strengthening the infocomm and media sector's collective response to cyber and technological threats arising from frontier AI models.
"Ultimately, the objective is to help organisations adopt AI with greater confidence," Mr Ng said.