Singapore Unveils Agentic AI Governance Framework and Updated Data Guidelines

Singapore authorities released three major AI governance publications between May and June 2026, covering agentic AI frameworks, legal responsibility for AI agents, and new guidelines on personal data use in generative AI. The documents, analysed in a Latham and Watkins client alert published this week, signal Singapore's accelerating push to codify AI governance as the technology moves beyond experimental deployment.

The Infocomm Media Development Authority (IMDA) published its Model AI Governance Framework for Agentic AI (Version 1.5) on 20 May, updated on 5 June. The framework targets organisations deploying autonomous AI agents — systems that independently plan, make decisions, and take actions over multiple steps. It organises recommended practices around four pillars: assessing and bounding risks upfront, maintaining meaningful human accountability, implementing technical controls and lifecycle processes, and enabling end-user responsibility. The framework identifies five harmful outcome categories — erroneous actions, unauthorised actions, biased outcomes, data breaches, and system disruptions — and highlights additional risks for multi-agent systems including agent sprawl, collaborative failures, and unpredictable emergent behaviours.

Alongside the governance framework, IMDA released a discussion paper exploring how legal responsibility should be allocated for AI agents — a question regulators globally are grappling with as autonomous systems make increasingly consequential decisions. Separately, the Personal Data Protection Commission (PDPC) published proposed advisory guidelines on the use of personal data in generative AI on 2 June, with public consultation closing on 14 July 2026. Together, the three documents provide the most comprehensive AI regulatory blueprint yet from a Southeast Asian jurisdiction.

Why it matters for Singapore: Singapore continues to differentiate itself through practical, principles-based AI regulation rather than heavy-handed rules. The agentic AI framework is among the first of its kind globally to specifically address the risks of multi-agent systems — areas where even major Western regulators are still developing guidance. For Singapore-based enterprises deploying AI agents in production, these frameworks provide much-needed legal clarity on accountability structures, data protection obligations, and risk management practices that will shape how the city-state's AI economy develops over the next several years.