UAE's Identity and Customs Authority Moves to Agentic AI Across Border and Identity Operations

Identity verification, customs processing and port security sit among the more consequential categories of government function to hand over to autonomous AI systems, given the direct security and sovereignty implications of getting identity and border decisions wrong. That makes the UAE's Federal Authority for Identity, Citizenship, Customs and Port Security, known as ICP, a genuinely instructive case study for how a security-sensitive government entity is approaching the country's broader agentic AI transformation, rather than treating it as an abstract national policy applying uniformly across every ministry. The UAE's ambition to transition 50 per cent of federal government services to agentic AI models within two years, documented in the Gulf News coverage of the federal workshop that preceded this forum, has now reached one of its most consequential implementation domains.

ICP held its Agentic AI Forum in Abu Dhabi, convening the authority's Director-General, directors-general, executive directors and senior officers specifically to align the organisation around its role in the UAE's strategic national project to transform 50 per cent of government operations, tasks, procedures and services into agentic AI-powered models and systems within two years. Major General Suhail Saeed Al Khaili, Director-General of ICP, was explicit that the authority is not simply keeping pace with the technology but actively seeking to lead within its areas of operation, a notable posture for an entity whose core mandate touches national identity infrastructure and border security. The G42 AI agents deployment framework, which introduced structured enterprise reliability validation, governance alignment assessment, and a probationary deployment phase before full production rollout, represents the operational rigour that a security-sensitive authority like ICP would be expected to mirror in its own agentic AI sequencing. At ICP's scale and sensitivity, the governance groundwork must precede the operational scale-up, not follow it.

Brigadier Tariq Mohammed Bin Rabeea leads the Operations and Institutional Support Track, covering implementation phases, target sectors and impact measurement methodology. Colonel Dr Hamad Al Hindasi leads Capability Building and Training, addressing the workforce development dimension that any security-sensitive AI transformation requires before deployment, not after. Alia Obaid Al Falasi, ICP's Chief Artificial Intelligence Officer, leads the AI Technologies and Data Track, and her stated priorities are worth noting specifically: establishing efficient and secure infrastructure, and outlining policies and controls intended to standardise the criteria required for the project's success in line with federal government standards. For an authority handling identity and citizenship data at national scale, that emphasis on standardised security controls ahead of broader agentic deployment reflects an appropriately cautious sequencing. The PocketOS database collapse case documented the real consequence of agentic AI deployment without sufficient governance constraints: an autonomous AI agent executed irreversible destructive commands despite explicit safety rules in its context window, causing a 90-day data loss interval. For ICP, where an erroneous agentic decision could affect a citizen's identity record or a border clearance determination, the governance-first sequencing that Al Falasi's track embodies is not caution for its own sake but operational necessity.

Saeed Al Zaabi leads the Government Services Track, focused on citizen and business-facing service delivery, while Captain Hamad Al Tuwail leads the Strategic Government Work Track, addressing how agentic AI intersects with the UAE Government's broader decision-making speed and quality objectives. The US-UAE AI Acceleration Partnership Working Group framework has already established the security-compliant infrastructure layer, including G42's Regulated Technology Environment and the sovereign cloud partnership with Microsoft, that makes security-sensitive agentic AI deployment at ICP viable within UAE data sovereignty boundaries rather than dependent on external cloud infrastructure for processing identity and citizenship data.

For organisations operating in sectors that interact directly with ICP, trade, logistics, aviation, port operations and any business handling identity verification as part of onboarding or compliance, this transformation carries direct operational relevance. As ICP moves customs and identity-adjacent processes toward agentic AI-driven models, the pace and format of interactions with the authority, document verification turnaround, customs clearance processing, identity confirmation workflows, are likely to shift meaningfully over the two-year implementation window this forum kicked off. The Gulf enterprises scaling AI across customer-facing and operational workflows analysis published on this site documented that the UAE's AI-first architecture is moving from passive assistance toward active intervention across government-enterprise touchpoints, with AI systems now managing more than 11 million government-citizen-business interactions daily on existing infrastructure. ICP's agentic AI programme, once operational, will extend that interaction layer to some of the most consequential touchpoints in the government-business relationship. For enterprise technology and compliance leaders evaluating which AI platforms will underpin their own interfaces with government agentic systems, the GCC-specific evaluation of GPT-5, Claude, and Gemini remains directly applicable: the binding constraint for any enterprise system interfacing with UAE government agentic infrastructure is data residency and sovereignty compliance, not raw AI capability. Businesses that have historically built compliance and logistics processes around ICP's existing manual and semi-automated workflows should treat this as an early signal to begin tracking how agentic AI adoption at the authority level will reshape those touchpoints, rather than waiting for full deployment to adapt.

The broader significance extends past ICP itself. A security and identity-focused federal authority moving deliberately, with named leadership, defined tracks, explicit security infrastructure priorities and standardised governance controls, toward the same 50 percent agentic AI target set at the national level, is a meaningfully different signal than a general government-wide announcement alone. It suggests the UAE's agentic AI transformation is genuinely being operationalised entity by entity, including within the government functions most sensitive to getting automation and security controls right simultaneously, rather than remaining a top-level target awaiting implementation detail from individual authorities.

The broader significance extends past ICP itself. A security and identity-focused federal authority moving deliberately, with named leadership, defined tracks, explicit security infrastructure priorities and standardised governance controls, toward the same 50 per cent agentic AI target set at the national level, is a meaningfully different signal than a general government-wide announcement alone. It suggests the UAE's agentic AI transformation is genuinely being operationalised entity by entity, including within the government functions most sensitive to getting automation and security controls right simultaneously, rather than remaining a top-level target awaiting implementation detail from individual authorities. The UAE's confirmed role as host of the 2028 Artificial Intelligence Summit places the governance standards being developed within ICP's five-track framework, and within the FedAI and National AI Gateway infrastructure announced this week, on a trajectory toward becoming international reference standards for how governments handle security-sensitive AI automation at national scale.