The United Arab Emirates has developed an artificial intelligence (AI) “control tower” designed to monitor and oversee AI agents deployed across its federal government. The initiative aims to ensure transparency and accountability as the country advances its broader strategy to integrate AI technologies into public services, Minister of State for AI, Digital Economy and Remote Work Applications Omar Al Olama stated during a panel discussion at the World Economic Forum’s Annual Meeting of Global Future Leaders in Dubai.
Under this system, every AI agent operating within the UAE federal government must connect to the control tower, which records all actions and decision-making processes of these AI entities. Both automated AI audits and human supervisors will continuously review this data to prevent agents from exceeding their authorized roles, Al Olama explained.
The development of this control tower accompanies a comprehensive federal framework approved by the UAE Cabinet in May to implement “agentic AI” across various ministries and government bodies. The government aims to transition at least 50 percent of federal services and operations to AI-based systems within two years. This framework includes the establishment of dedicated teams per government entity and performance metrics to evaluate the transformation’s impact.
Complementing these efforts, the UAE has launched a large-scale training program involving 80,000 federal government employees to prepare the workforce for AI integration. Simultaneously, the government is working on the Federal Agentic AI Technology Platform (FedAI), which will provide a unified infrastructure for AI development and deployment, alongside a National AI Gateway to facilitate secure access to multiple AI platforms, including commercial and open-source models running on sovereign infrastructure.
To determine appropriate AI adoption, the government conducted an extensive assessment of federal services. Approximately 70 percent of these services were deemed to have a tolerance for AI deployment, while the remaining 30 percent were viewed as carrying risks exceeding acceptable thresholds. “We’re not looking at the 30 percent where the risk of making a mistake is higher than our tolerance,” Al Olama noted.
The approach underscores the UAE’s emphasis on safeguarding human livelihood. Al Olama stated that any AI decisions that could significantly impact an individual’s livelihood represent a strict “red line,” beyond which autonomous AI decision-making is not permitted. The government continues to evaluate AI applications based on the potential harm of incorrect decisions, with lower-risk functions such as certain document renewals allowing for greater AI autonomy and higher-risk areas subject to more rigorous oversight.
Gabon’s Minister of Digital Economy and Innovation, Mark-Alexandre Doumba, participating in the same panel, highlighted the necessity for governments to establish clear regulations both within AI technologies and through international policy frameworks. He emphasized the importance of embedding safeguards in AI coding and standards that define permissible agent behaviors and decisions requiring human involvement.
The UAE’s work reflects a broader global dialogue on balancing AI innovation with ethical constraints and human-centered governance. The government acknowledges that the full capabilities of agentic AI are still emerging and insists that human life and livelihoods must not be reduced to monetary considerations as these technologies evolve.