The UAE has drawn a red line on the use of artificial intelligence (AI) in government: no AI agent will be allowed to make decisions that could affect a person’s livelihood.
Omar Sultan Al Olama, Minister of State for Artificial Intelligence in the Office of the Prime Minister, said the rule applies “especially at this point, because we don’t know the true capabilities of technology”. “We don’t think that human life or human livelihood is something that has a dollar figure value to it. It’s invaluable,” he said.
He was speaking at a panel discussion titled Trust Test for Agentic AI at the World Economc Forum’s Annual Meeting of Global Future Leaders in Dubai on Wednesday, where he also took questions on global regulation, public fears over data centres, and job losses.
Stay up to date with the latest news. Follow KT on WhatsApp Channels.
The rapid rise of artificial intelligence has sparked global concern about its impact on jobs, with millions of workers fearing automation could replace their roles. While AI promises greater productivity and efficiency, experts warn that routine jobs face growing risks, highlighting the urgent need for reskilling, education and workforce adaptation.
Al Olama said UAE leaders have set a target for AI agents to deliver 50 per cent of federal government services within two years.
Officials mapped all government services and found that around 70 per cent could tolerate some use of AI. In the remaining 30 per cent, he said, the risk of an error exceeds the government’s tolerance, so these are not being considered for automation.
Low-risk tasks such as renewing a passport are among those suited to agents, he said. Services where an agent’s actions would need a higher standard of scrutiny are excluded for now.
AI agents are audited
Al Olama also disclosed what he suggested was the first public mention of an “agent control tower” on the federal government side. Every agent must plug into it, and every action and “thought” is logged.
He said AI agents are audited continuously by both other agents and humans.
Government agents are also deployed only by what he described as a responsible UAE party that understands the ramifications, rather than bought in from companies that do not.
On global rules, Al Olama said the debate should not be about slowing down or speeding up AI. “That train has left the station,” he said, citing the US president’s stated determination not to lose a race with China.
Instead, he urged countries to agree first on the most extreme risks, starting with autonomous weapons that can decide to kill. He said there is little debate that such decisions should not be left to machines, yet no global consensus exists.
“If you sign up to it, and your adversary doesn’t, it puts you at a disadvantage,” he said, comparing the effort to decades of attempts to curb nuclear weapons.
He called for governments to set a firm deadline for global AI regulations. Having served in his post as AI minister for nine years, he said he did not want to look back nine years from now and find no progress.
He said warnings that ChatGPT-2 was too dangerous for public release came in the early 2020s. Since then, he said, the UAE has taken part in discussions at the G20, G7 and the UN without reaching consensus.
Fear, jobs and data centres
Al Olama said fear comes from not understanding a technology. The UAE has focused on education, including an annual AI Day on October 29 when every resident receives an SMS with a starter prompt.
He said data centres in the UAE do not raise energy prices because the country uses solar and nuclear power and has spare capacity. He added that the UAE has no natural water to draw from the grid and relies on desalination.
On jobs, he said a programme to train one million people as “super prompters” will conclude this month. Every federal employee is due to complete extensive AI training by August next year, with half already done.
He said the UAE model is unique and would not map directly onto the US, China or India, but parts could be scaled.