OpenAI said at the ATx Summit that it will open its first Applied AI Lab outside the United States in Singapore under a new initiative called OpenAI for Singapore, backed by a commitment of more than S$300 million. The lab will position Singapore as an in-region engineering and deployment centre and make the city-state one of OpenAI’s global hubs for forward‑deployed engineers who will work directly with organisations on AI deployment and integration. The investment is expected to create more than 200 Singapore‑based technical roles over the next few years.
As part of the initiative, OpenAI outlined a programme of local collaborations focused on deployment and talent development. Planned activities include work with government agencies and partners on education and workforce programmes inside the Ministry of Education and GovTech, the launch of a Singapore chapter of the OpenAI Academy, participation in the National AI Impact Programme, and Codex for Teachers hackathons aimed at upskilling educators and developers.
Concurrently, Singapore’s Infocomm Media Development Authority (IMDA) published an updated governance framework for agentic AI, a revision of its guidance originally launched at the World Economic Forum in January 2026 and building on the 2020 Model AI Governance Framework. IMDA said it revised the document after gathering feedback and case studies from more than 60 organisations, including AWS, DBS, Google, and Salesforce, to reflect practical lessons from real deployments.
The updated framework adds guidance on risks tied to multi‑agent systems, third‑party agents, automation bias, and human accountability, and now includes more than ten case studies demonstrating application of the recommendations. IMDA listed case study contributors such as Ant International, City Developments Limited, Dayos, Google, OCBC, PwC, Tencent, Workday and GovTech Singapore, providing exemplars for governance and controls.
Those case studies supply concrete controls and rollout patterns for builders. Dayos described an AI ticketing agent that uses tiered risk levels: low‑risk, reversible actions are handled automatically with biweekly audits; moderate‑risk actions require human approval; and high‑risk actions are excluded. Tencent’s CodeBuddy is reported to plan, write and deploy code with access to filesystems, terminals, external APIs and tools, while enforcing preset defaults, configurable permissions and mandatory human approval for file edits, shell commands, network requests or suspicious commands.
GovTech’s rollout of agentic coding assistants began with an initial phase limited to its own employees.
Sources
Replies (0)
No replies in this topic yet.
