The UAE has made a decision that no other country has made at this scale. By 2028, half of its federal government operations will run on artificial intelligence systems that can execute tasks and make decisions without a human pressing a button at every step.
Sheikh Mohammed bin Rashid Al Maktoum, Vice President and Prime Minister of the UAE, announced the target in April 2026. The country aims to become the first government globally to operate at this scale using autonomous systems. The plan covers government sectors, services, and operations. Performance evaluations for ministers and federal entity leaders will now include how quickly they adopt these systems.
On September 8, 2026, the Ministerial Council for Artificial Intelligence and Development met at Qasr Al Watan in Abu Dhabi to review the second phase of the project. Sheikh Mansour bin Zayed Al Nahyan, Vice President, Deputy Prime Minister and Chairman of the Presidential Court, chaired the session. The Council confirmed that the first phase, which assessed government services and set priorities, is complete. Projects submitted by ministries and federal entities will now move into implementation using agentic AI.
The central piece of infrastructure for this rollout now has a name. FedAI is a national technical ecosystem that will allow federal entities to build, adopt, and operate agentic AI solutions within a secure framework. The Council reviewed its plans alongside a customer experience lab designed to apply agentic AI across government services and internal competitions to engage employees in building solutions.
Agentic AI Does Not Just Answer Questions
Most people know AI as a tool that responds to prompts. You ask a question, and it generates an answer. Agentic AI works differently. It takes actions. It can monitor systems, identify when something needs to change, execute a series of steps to address it, and report back. The system operates with a level of autonomy that goes beyond answering questions.
Sheikh Mohammed framed the shift directly. He said AI is no longer a tool. It analyses, decides, executes, and improves in real time. It will become an executive partner to enhance services, accelerate decisions, and raise efficiency.
The scale of this ambition becomes clearer when you look at the seven pillars the project covers. These are strategy and projects, foresight and strategic intelligence, policies and governance, government performance, global competitiveness, and innovation in government work. These are not simple administrative tasks. Policy formulation and strategic foresight require judgment, context, and an understanding of long-term consequences. The UAE is asking AI systems to assist with those functions.
The Infrastructure That Makes FedAI Necessary
A target of 50 percent adoption in two years creates a coordination problem. If each ministry builds its own AI systems in isolation, the result is fragmentation. Different tools, different data formats, different security standards. None of it compounds.
FedAI exists to solve that problem before it happens. The national technical ecosystem gives federal entities a shared set of standards and tools for building and operating agentic AI. Rather than each ministry starting from scratch, FedAI provides the foundation.
The project also includes a customer experience lab and internal competitions to engage employees in developing solutions. These initiatives matter because agentic AI adoption is not just a technology project. It requires changing how government employees work and how services get designed.
Human Oversight Is Not Optional
The UAE has built its entire approach around a principle called human leads, AI enables. This language sets a boundary. Human officials lead. AI provides analytical power, processes information at speed, and surfaces options. The final call belongs to people.
The first phase of the project focused on building government capabilities and assessing which services, operations, and tasks are suitable for AI transformation. The implementation phase will classify tasks and coordinate adoption across entities. Not everything will be automated. High-stakes transactions involving financial payments, legal rights, or sensitive personal data require strict human-in-the-loop safeguards, according to experts consulted on the project.
AI and data expert Vasudha Khandeparkar described how the question is shifting. She said the focus is moving from which jobs stay human to which decisions should remain human. Accountability, ethical judgment, and navigating ambiguity remain fundamentally human responsibilities.
The Security Question Grows With Autonomy
Moving from AI that answers questions to AI that takes actions changes the security challenge. Debo Zhang, CEO of HONOR GCC, explained that as AI transitions from reactive chatbots to autonomous agentic execution, the attack surface expands significantly. Malicious actors will increasingly use adversarial attacks and deepfakes to manipulate these agents.
Zhang also pointed out that securing government infrastructure alone is not enough. The devices citizens use to access AI-enabled services form part of the security chain. A highly secure government AI network is only as resilient as the personal devices citizens use to interact with it.
The UAE has already taken steps to build institutional backing for these concerns. In June 2026, the government merged its Office of Artificial Intelligence, Digital Economy and Remote Work Applications with the UAE Data Office to create a single national authority consolidating public data, AI, and digital government capabilities. This authority gives the design guide and governance frameworks an institutional home with genuine authority over both the data infrastructure and the AI systems.
READ ALSO:
UAE Launches First AI Learning Module for National Experts Programme
How the UAE Is Using Smarter AI to Fix Its Food Industry Bureaucracy
The UAE is training 25,000 teachers in AI as part of a national strategy
A Design Guide That Sets Actual Rules
The UAE published a unified design guide for agentic AI-powered public services in August 2026. The guide moves the project from stated ambition into specified behaviour. It sets binding design principles rather than general aspirations.
Two principles stand out. First, an agentic AI system should never ask a user to provide information they have already submitted previously. This sounds simple. Delivering it requires genuine cross-entity data interoperability. Information submitted to one government body needs to be accessible, with appropriate authorisation, to any other agent handling a related request.
Second, users must be kept informed at every step of their service journey. This transparency requirement pushes against the risk of an autonomous system operating as an opaque black box between request and outcome.
The guide also anticipates situations where a citizen needs to interact with an actual person. It codifies human handoff points into the system design. System designers building agentic services now work from a specified requirement for human involvement, not just a stated intention.
The Implication for Government Workers
The human element of this transformation extends beyond oversight. The UAE Cabinet approved the largest government training program in the country’s history. The goal is to provide 80,000 public sector workers with agentic AI tools and abilities. Personnel from all ministries, authorities, and government organisations will receive training.
Sheikh Mohammed described the intent directly. He said the country is investing in its people. Every federal employee will be trained to master AI, building one of the world’s strongest capabilities in AI-driven government.
This approach reflects a recognition that agentic AI does not eliminate the need for human workers. It changes what those workers do. Rather than processing forms or handling routine requests, employees will design solutions, supervise systems, and manage exceptions. Huda Al Hashimi, Deputy Minister of Cabinet Affairs for Strategic Affairs, said the workshop marked the practical start of deploying AI models to strengthen government performance.
The Timeline Is Aggressive, and the Stakes Are High
Converting half of government operations to autonomous AI systems in 24 months is not a typical pace. Most governments that announce AI integration targets work on longer timelines. The UAE’s compressed schedule suggests significant prior planning and substantial investment in technology infrastructure.
The decision to tackle all seven pillars simultaneously, rather than piloting in one area first, amplifies the complexity. Policy work and strategic foresight are inherently harder to automate than data entry or routine approvals. The framework acknowledges this by including structures and governance as a dedicated pillar, suggesting the government recognises it may need to reshape how agencies organise themselves to make AI integration viable.
Sheikh Mansour described the target clearly. He said the goal is to transform AI capabilities into tangible impact on government efficiency, quality of services, and speed of execution, and to build national capabilities that can lead this transformation efficiently and responsibly. He also said the UAE is building a government system that is smarter, more flexible, and more proactive, one that uses agentic AI to shape services, operations, policies, and decisions within secure digital infrastructure.
The world will watch whether this works. If the UAE succeeds, it will have demonstrated that governments can operate at a scale and speed that previously seemed impossible. If it struggles, the challenges will reveal the limits of autonomous systems in the complex, high-stakes work of governing. Either way, the experiment has already begun.
