Josephine Teo urges nations to proactively address agentic AI governance risks

At the AI Summit in Delhi, Singapore's Minister for Digital Development and Information Josephine Teo warned that the autonomous nature of agentic AI introduces new risks and that countries must govern them proactively, especially in healthcare and social services. Government touchpoints with citizens are sensitive — getting a health, social-security or benefits answer wrong, and then having a citizen act on it, can do serious damage.

She called for stronger international cooperation to turn principles like transparency and fairness into practical safeguards, with standardised evaluation methods that work across regulatory contexts. Capacity building matters too, so all countries can engage with the technical evidence, not just those with large AI research ecosystems. Singapore is actively testing, developing standards and working with the private sector to verify safe and reliable AI use.

Teo warned that regulation alone cannot resolve AI's wider social strains. Expecting AI rules to solve inequality is unrealistic — societies need what she called social solidarity, including support for job transitions, housing, healthcare and education. In a fragmenting global tech landscape, Singapore's strategy is to remain a trusted node: connected to cutting-edge technology while keeping misuse risk low.

Caption language: en · Fetched: 2026-05-02

Singapore's digital development and information minister Josephine Tuo has urged countries to be proactive about governing the emerging risks in agentic AI especially in areas like healthcare and social services. That's because of the autonomous nature of such tools that may introduce new risks. Speaking at the AI summit in Delhi, she says the consequences also become complex and unpredictable. Government is high-risisk because the touch point with citizens are very sensitive. No citizen and no government wants to um make serious mistakes when they interact with their citizen. telling them uh things about their health, telling things about their social security, telling them about things to do with their um uh benefits that that are not accurate and having them not just being told but acted upon. Mrs.

Tio says Singapore is actively testing, developing standards, and working with the private sector to verify and build trust in the safe and reliable use of AI. She says such work cannot be done alone and Singapore welcomes more cooperation with other partners. Mrs. Ste also called for closer collaboration to turn AI safety principles into practical safeguards amid fragmented global governance. She says the challenge is turning concepts like transparency and fairness into practical actions and this will require standardized evaluation methods that work across different regulatory contexts. We need capacity building so that all countries can meaningfully engage with the technical evidence and not just with the large AI research ecosystems.

I would encourage all stakeholders to view scientific input not as a constraint on policy flexibility but as a foundation for more durable effective governance that can maintain public trust. The digital development and information minister also highlighted the importance for countries to work together to use AI to mitigate risks and improve safety. She said this at a panel discussion on an international AI safety report published earlier this month. Now the report focused on generalpurpose AI which are models and systems that perform a wide range of tasks. These include writing code, analyzing data, generating images, and answering complex questions. The most capable of these systems are at the frontier of AI development. The report also listed risks including malicious use of scams and cyber attacks.

They also malfunctions where systems fabricate information, produce flawed outputs, or behave unreliably. Another systemic risk is that it can cause market disruptions and affect people's ability to make informed choices. The report also noted that there is no single solution to manage these risks. It says multiple safeguards are needed including threat modeling, capability testing, and incident reporting. It also pointed out governance gaps where rapid development and limited transparency make oversight difficult. Ultimately, we should try and get to a point where the end user has an assurance of safety that they don't have to be thinking so hard about whether you know the proper tests have been applied. We're not there yet. Uh but I think we need to find a way you know to work out the road map. Mrs.

Tio says Singapore's approach to AI in an increasingly fragmented global tech landscape is to remain a trusted node. This means keeping Singapore connected to cuttingedge technology while ensuring it is not abused and the risk of misuse is minimized. She points to growing risks of technology decoupling, saying that Singapore stays trusted by acting in a consistent principled way, but she warns that regulation alone cannot deal with AI's wider social strains. I think sometimes there is a um um a desire uh a tendency to want to think of uh ways of regulating AI um in order to slow down its um its its u advance and perhaps you know to uh try and forall the risk. Uh I'm not underestimating the need uh for example in making sure that there are guardrails on AI safety.

I I just want to say that that those are important, but to overexpect AI regulations to deliver on the other uh important issues such as the potential for greater social inequality. I think this is unrealistic. Instead, Mrs. stuosis societies need to strengthen what she terms social solidarity and this includes support for job moves and access to housing, healthcare and education to sustain trust as AI spreads.