OpenAI has announced that its first Applied AI Lab outside the US will be set up in Singapore as a new collaboration with the Ministry of Digital Development and Information.
Under the project called OpenAI for Singapore, OpenAI has promised an investment of more than S$300 million in the initiative. It was announced at the ATx Summit.
The lab will be involved in generating more than 200 technical jobs based out of Singapore in the coming years. OpenAI added that Singapore would also be one of its engineering hubs to deploy engineers across companies and help them deploy AI technology. OpenAI stated that it plans to align the lab with the AI Mission priorities of Singapore, which include public service, finance, and digital infrastructure.
Consider deployment and recruitment of talented individuals
These will include collaborations on educational and workforce development initiatives with the help of government bodies and local partners within the Ministry of Education and GovTech. The OpenAI is also looking to collaborate on supporting educators by having the OpenAI Academy Singapore Chapter, joining the National AI Impact Programme and conducting Codex for Teachers Hackathons.
The collaboration entails working together with local partners for conducting accelerator programmes for AI native startups in the form of workshops for micro entrepreneurs and small businesses on how they can utilize AI for themselves and their customers.
Chng Kai Fong, the Permanent Secretary for Digital Development and Information, mentioned that Singapore’s approach towards AI involves developing new industries, building global frontier companies, and developing appropriate skill sets among its workers.
Also Read : United States Government Expands AI Suppliers and Reassesses Anthropic’s Role
Singapore modifies agentic AI governance system
Singapore also has improved the framework that was introduced for the governance of agentic AI by the Infocomm Media Development Authority in the World Economic Forum held in January 2026. The framework draws on Singapore’s earlier Model AI Governance Framework for AI, which was released in 2020. It gives guidance to the organizations about using agentic AI in a responsible way that minimizes risks of agentic AI.
Now, the framework has been upgraded after feedback and case studies collected from the industry, where more than sixty organizations such as AWS, DBS, Google, and Salesforce provided input for the upgrading of the framework.
The new guidance relates to issues regarding risks associated with multi-agent systems, risks with third-party agents, automation bias, and human accountability in cases of agentic AI. More than ten examples of implementation of the guidance have been added to the framework in case studies.
These case studies have been made possible by contributions from Singaporean and international organizations such as Ant International, City Developments Limited, Cyber Sierra, Dayos, Google, Knovel, OCBC, PwC, Stability Solutions, Tencent, Terminal 3, Workday, X0PA, and GovTech Singapore.
Also Read : Autonomous AI Systems Are Testing Governance in the Physical World
Case studies have proved that governance exercises control
A specific case study revolves around the activities of Dayos, a Singapore-based AI automation company operating in the United States. Specifically, Dayos created an AI-powered ticketing agent that processes internal IT requests. The system was capable of resolving some requests automatically or transferring requests for human resolution, when appropriate.
The decision about the actions taken by the agent was determined based on risk levels. Low-risk or easily reversible tasks, such as password changes, were handled automatically but underwent audit every two weeks. Moderate-risk actions were allowed to be performed by agents, but a human operator had to give permission first. High-risk actions, which had limited ability to reverse effects, were not included in the list of tasks to be automated.
Another case study comes from Tencent and involves the development of CodeBuddy, which is a code automation tool developed by Tencent Cloud. The software was capable of planning, writing, and deploying computer code according to user instructions and could access file systems, shell commands, external APIs, and other MCP utilities.
The software utilized certain default settings and permissions. Any action that involved file modifications, shell command execution, network access, and any use of external programs would require human approval.
It translates complicated commands into simple English before the user gives consent to the system. Even when some other commands have already been pre-approved, suspicious ones will still need human confirmation.
(Photo by Mike Enerio)
