UAE leads on AI agents, but oversight lags, survey finds

The speed with which UAE organisations have deployed artificial intelligence agents is now among the highest recorded anywhere, but a new industry survey suggests that the oversight mechanisms needed to govern those systems have not kept pace with adoption.

The findings come from the 2026 Global AI Confessions Report, CIO Edition, commissioned by the data science and AI platform company Dataiku. The analysis draws on responses from 685 chief information officers across the UAE, the US, the UK, France, Germany, Japan, South Korea and Singapore. It places UAE enterprises at the top of the table for agentic AI deployment while flagging significant weaknesses in containment and accountability practices.

The deployment numbers are striking. According to the report, 62 per cent of UAE chief information officers say their organisations run more than 50 AI agents, and 15 per cent report operating more than 500. That is the highest proportion across the eight markets surveyed and substantially above the 9 per cent global average for the larger figure. AI agents, digital personal assistants powered by artificial intelligence to carry out repetitive tasks, have become increasingly prevalent across sectors in recent years, and technology companies have claimed they can match and often outperform humans in many functions, freeing employees for other work.

The hiring picture reinforces that momentum. Among CIOs whose organisations are creating or planning embedded roles for AI experts or forward engineers, 22 per cent of UAE respondents expect to recruit more than 10 people over the coming year, the highest figure of any country in the study.

Yet the survey’s governance findings point in a different direction. Only 5 per cent of UAE CIOs say their organisation could reliably identify and contain a problematic agent across environments within one to two hours, among the lowest results in the study and half the worldwide figure of 10 per cent. The poll also found that 27 per cent of UAE chief information officers said they could intervene later in the day to address a hypothetical issue with a rogue AI agent, meaning two thirds of those surveyed felt they could not contain the agent on the same day. The analysis notes this comes after incidents in which AI models escaped containment and compromised the digital infrastructure of outside entities.

Kurt Muehmel, head of AI strategy at Dataiku, urged perspective when reading the results. “UAE organisations already do one important part of this well, with 80 per cent of UAE CIOs having seen an agent act against business intent or policy while still operating within its technical parameters, but 62 per cent say they caught the problem before it caused harm, the best result globally,” he said.

Muehmel also set out steps that businesses and organisations in the UAE could take to strengthen their approach. “Keep one inventory of every agent, whoever built it and whatever platform it runs on,” he said. He added that an owner should be designated for every agent and that a decision must be made in advance about who would have the authority to stop it. Organisations should also “standardise how agents are tested, changed and retired … and practise containment the way you practise a cyber security incident response.”

Taken together, the results can provide a road map for making the UAE’s approach to AI more robust, pairing rapid deployment with clearer lines of responsibility and faster containment capability.

Sid Bhatia, Dataiku’s vice president for the Middle East, Turkey and Africa, said the overall analysis bodes well for the country’s push to be a leader in AI and in the use of AI agents. “The UAE has established a strong foundation for agentic AI, with enterprises deploying at scale, investing in specialist expertise, and preventing a high proportion of agent incidents from causing real-world harm,” he said. Whether the containment gap narrows as quickly as deployment has widened remains the open question for the year ahead.