OpenAI opens AI lab in Singapore as IMDA updates AI framework

OpenAI Opened the first Applied AI Lab in Singapore. This lab is part of a new partnership with the Ministry of Digital Development and Information.

This initiative called OpenAI for Singapore has been announced at the ATx Summit and funded over $300 million.

We will create more than 200 technical jobs based in Singapore for the next few years. OpenAI said that Singapore will be one of the global bases of forward-、 engineers working with organizations in AI deployment. OpenAI said that this laboratory’s efforts will be in line with the priorities of Singapore’s AI mission, including public services, financial and digital infrastructure.

Focus on Introduction and Human Resources

The company cooperates with government agencies and local partners with educational and labor programmes in the Ministry of EducationHomevTech OpenAI will also support educators through the Singapore chapter of OpenAI Academy, participate in the national AI impact program and conduct Codex for Teachers Hackathon.

This partnership includes a plan to work with local partners in the Accelerator Program for AI Native Startups in the form of workshops for small entrepreneurs and small businesses to cover how founders and small businesses can utilize AI in their operations and customer services.

Mr. Chng Kai Fong, head of Digital Development Information, said that AI’s response to Singapore includes new field growth, global frontier firm fixation, and grant related skills to workers.

Singapore Updates Agent AI Framework

Singapore also updated its agent-type AI governance framework, which was announced by Infocomm Media Development Authority at the World Economic Forum in January 2026. This framework is based on the AI governance framework for emerging AI in Singapore introduced in 2020 and provides organizations with guidance on responsible implementation of AI agents, including measures to mitigate the risks intra-agent AI.

IMDA will update the framework after seeking feedback and case studies from the industry,,DBS,Google MapWe have created a revised version that reflects opinions from more than 60 organizations, such as Salesforce.

This update includes guidance on multi-agent systems, third-party agents, automated bias, and risk related to human accountability. This framework now contains more than 10 case studies that show how organizations applied that recommendations.

Case Study was contributed by Singapore and international organizations includingt International, City Developments Limited, Cyber Sierra, Dayos, Google,novel, OCBC, PwC, Stability Solutions, Tencent, ter。 3, Workday, X0PA, and GovTech Singapore.

Case Study shows governance management

One of our case studies focuses on Dayos, an enterprise AI automation company headquartered in Singapore. Dayos has built a ticketing agent that uses AI to handle internal IT requests. Agents can automatically resolve some requests and route requests to humans as needed.

Dayos decided to use step-by-step risk levels to determine which actions the agent can perform. Actions that can be restored at low risk, such as resetting passwords can be automated and audited in a biweekly basis, but middle-risk actions require human approval before execution. More risky actions, such as changing reversible rights, were excluded from the agent’s permissions.

Tencent contributed to the case study of CodeBuddy, an agent AI coding system developed by Tencent Cloud. CodeBuddy can plan, create, and deploy code through natural language commands, file systems, terminal commands, and external APIand MCP Access tools

CodeBuddy uses pre-config。 defaults and configurable privileges. Actions such as editing files, running shell commands, running network requests, and using external tools require human approval.

The system can be written in plain words before the user approves complex commands. Even if the same command has been approved in advance, suspicious commands still require human approval.

GovTech Singapore’s case study covers the introduction of agent coding assistants in the government. The first phase included GovTech employees, the external tools were not allowed and included a low-risk system. GovTech has developed a framework for connecting central log records and approved external tools. The agency also tested the system against potential .

(Photo courtesy of Mike Enerio)

Related Items: GPT-5.5 is the most competent agent AI model ever for OpenAI

Want to learn more about AI and Big Data from industry leaders? Check out our AI & ポ Expo in Amsterdam, California and London. This comprehensive event is part of TechEx and is located in the same location as other major technology events. Click here for more information.

TechForge Media Learn more about other enterprise technology events and webinars that will be held in the future.