The United Arab Emirates has set a two-year deadline to move half of its federal government operations to agentic AI models, making it the first state to confront a question no government has yet answered publicly: which decisions may a machine make on its own? More than 100 federal officials gathered this month in a workshop to begin classifying government tasks into those an AI agent may execute and those it may only recommend.
The workshop, organised by the National Committee for the Agentic AI Project, launched the strategic track of the programme and set implementation priorities and timelines. The classification exercise is the core challenge. The UAE already has the infrastructure most governments are still procuring - digital identity, sovereign cloud and data-sharing layers - along with a performance system tracking over 150 million data points a month. What it lacks, and what no government has published, is a defensible method for drawing the line between machine-executed decisions and machine-recommended ones.
Two framings of one project
The project has been framed two ways. When Sheikh Mohammed bin Rashid Al Maktoum announced it, he said the UAE would become the first government to largely deploy agentic AI models - systems that manage operations and run independent series of actions without human intervention. The implementation language is different: "AI leads, human supervises."
Both framings can be true across a portfolio of thousands of tasks, but the gap between them is exactly what the classification framework must close. Huda Al Hashimi, Deputy Minister of Cabinet Affairs for Strategic Affairs, described the workshop as the operational beginning of deploying AI models across government activity. The programme runs across seven pillars, including policies, structures and governance, foresight and strategic intelligence, government performance, and training.
Until those frameworks are published, legal responsibility for an agent's decision remains unresolved. No government has fully answered it. The difference is that the UAE will have to answer it first, at scale, on a two-year clock.
Scale and training
Oversight sits with Sheikh Mansour bin Zayed Al Nahyan, with a taskforce chaired by Mohammad Al Gergawi, Minister of Cabinet Affairs. In May, the Cabinet approved the first package of government services to move to agentic AI, alongside the largest training programme in UAE government history: 80,000 federal employees, from ministers to new joiners. A national AI healthcare policy and digital records framework passed in the same session.
A Dubai training event in June drew more than 300 participants from 50 federal institutions, working toward a goal of identifying which services could be automated within 90 days. Another session that month, run by the Presidential Court and the Ministry of Cabinet Affairs, put 600 federal staff through the same material. For government professionals facing similar transitions, understanding how to evaluate AI readiness is becoming a core skill - resources like the AI Learning Path for Policy Makers address exactly this kind of governance and decision-making challenge.
Why the UAE's timeline matters beyond the Gulf
Governments across Asia and Europe are approaching the same decision from a standing start - most with an AI strategy, a national body, and no number attached to any of it. The UAE has attached a number and a date. Within two years, there will be public evidence about what agentic AI in government does to service delivery, headcount and accountability.
No other government has to move at that pace, but they will have to decide what they do with the results. The task classification frameworks are the documents to watch. When published, they will be the first serious attempt by any state to set out which decisions a government is prepared to let a machine make on its own. Everything else in this programme is procurement and training. That part is constitutional.
Why this matters for government professionals
If you work in government, the UAE's classification framework will become a reference point for your own agency's AI policy. The question is not whether AI agents will be deployed in your department - it is who will decide what they are allowed to do, and how that decision gets documented. The UAE is building that documentation under public scrutiny and on a deadline. The frameworks it publishes will likely become the template for every other government, including yours. For those who need to prepare for these decisions, AI for Government training can help build the foundation for evaluating what belongs in each category.
Your membership also unlocks:
