Human-Guided AI Drives UAE’s Next Phase in Combating Financial Crime.
Abu Dhabi: Drones monitoring ships, algorithms analysing invoices, police tracking cryptocurrency transactions and artificial intelligence screening more than 25,000 entities — across the UAE, advanced technology is increasingly becoming part of efforts to protect the financial system from crime.
The focus is now shifting from testing AI through proof-of-concept projects to deploying scalable solutions that can deliver practical benefits, according to Mahmoud Alsalah, Partner and Financial Crime Compliance Leader at PwC Middle East.
Alsalah made the remarks during a session hosted by the UAE General Secretariat on the third day of the 15th United Nations Congress on Crime Prevention and Criminal Justice in Abu Dhabi.
During the session, representatives from UAE Customs, Abu Dhabi Police, the Ministry of Economy and Tourism, and PwC Middle East highlighted practical examples of how artificial intelligence is being deployed across the country. They outlined how the technology is helping authorities screen shipments, track cryptocurrency transactions and better target inspections based on identified risks.
Drones monitor ships and containers
At customs checkpoints, the shift towards technology-driven inspections is already evident across the UAE’s borders and ports.
Maryam Amer, Director of the Anti-Money Laundering and Countering-Terrorist Financing Department at the Federal Authority for Identity, Citizenship, Customs and Port Security, said customs inspections have evolved significantly from conventional methods.
She explained that inspections now incorporate modern technologies across land, sea and air borders.
Drones are being deployed to inspect vessels and shipments arriving through maritime and air routes. Artificial intelligence also helps customs officers develop risk profiles and identify potentially suspicious activity, including unusual invoice values and irregularities linked to the origin of shipments.
Cross-border cash declarations are another area where AI is being used. Since large movements of cash across borders can carry money-laundering risks, AI-powered systems can generate alerts for customs officers reviewing declarations and help identify cases that may require closer examination.
Amer said these technologies also help reduce the time and effort required by customs officials during the clearance process.
Tracking cryptocurrency and virtual training
For Abu Dhabi Police, artificial intelligence is increasingly being integrated into efforts to maintain public safety and security.
Major Ali Alnuaimi of Abu Dhabi Police said AI is being used across three key areas: crime prevention, combating criminal activity, and training and developing the skills of police officers.
These efforts build on the UAE’s national threat assessment, which included virtual asset regulation and examined risks associated with fraud and scams.
As part of this approach, Abu Dhabi Police uses an analytical system to track financial assets, including cryptocurrencies and other virtual assets. The system maps the movement of these assets into and out of the UAE, helping authorities identify potentially suspicious financial flows.
Police also operate systems that monitor areas considered to have a higher risk of criminal activity, allowing authorities to focus resources and attention where they may be needed most.
AI is also playing a growing role in police training. Abu Dhabi Police uses virtual crime simulations to help officers practise identifying, combating and disrupting criminal activity, including offences involving emerging technologies.
Alnuaimi said the objective is to anticipate new risks rather than simply respond after crimes occur. By creating different scenarios, officers can prepare for evolving criminal methods and develop strategies to counter them.
He also highlighted the importance of cooperation between government agencies, describing the UAE’s approach as an interconnected system in which each authority supports the work of others.
For example, when customs officials identify a suspicious case, it can be referred to the police for further investigation and law enforcement action.
Data volumes too large for manual screening
For the Ministry of Economy and Tourism, one of the biggest challenges is the sheer scale of supervision required.
Counselor Salem Ahmed Al Tunaiji, Director of the AML/CFT Department at the Ministry of Economy and Tourism, said his team oversees around 25,000 entities each year, making it difficult to inspect every organisation using human resources alone.
To address this challenge, the ministry has adopted a risk-based assessment system supported by artificial intelligence, allowing officials to identify entities that may require greater scrutiny.
The technology supports an annual national risk assessment comprising around 300 questions answered by private-sector entities. The responses are analysed and used to guide both inspections and training programmes.
Al Tunaiji said the ministry needs to prioritise certain companies from the roughly 25,000 entities under its supervision. Responses from the assessment are therefore combined with information collected over the previous five years to help determine inspection priorities and identify possible indicators of suspicious activity, including risks involving cryptocurrencies and other virtual assets.
The findings are also used to shape training programmes. Sector-level information is circulated within the ministry, while some findings are shared with the General Secretariat.
The ministry’s next step is to develop an AI-powered mechanism capable of verifying documents submitted as part of regulatory controls and inspections. Given the volume of material involved, reviewing all documents manually would be difficult.
Keeping humans involved in AI decisions
While regulators and financial institutions continue to assess how much autonomy AI systems should have, Mahmoud Alsalah, Partner and Financial Crime Compliance Leader at PwC Middle East, said there is broad agreement on the need for human oversight.
This means that even when AI is used to analyse information or identify potentially suspicious activity, people should remain involved in critical stages of the process, such as preparing and submitting suspicious activity reports.
Alsalah also highlighted the limitations of artificial intelligence, particularly when it comes to establishing the true beneficial ownership of assets or companies.
He explained that AI may not be able to independently determine the actual beneficial owner, but it can rapidly analyse vast amounts of information, including Know Your Customer (KYC) records and phone numbers associated with an individual over a period of time.
This capability is particularly important because criminals can exploit information gaps between different institutions. Alsalah said the volume of available data is simply too large for a person to process manually, while organisations operating in isolation may fail to detect relationships that become visible when multiple datasets are analysed together.
Three ways AI is being used across government and business
Alsalah identified three main areas in which artificial intelligence is currently being applied across the public and private sectors.
The first — and currently the most widely used — is data collection and summarisation. AI can gather information from multiple sources and condense large documents, allowing investigators and analysts to spend less time compiling material and more time examining cases.
The second is identifying trends and patterns. AI can analyse large volumes of national and international data to detect emerging financial crime trends and typologies. In areas such as customs and cross-border movements, this can help authorities understand new methods and scenarios being used by criminals.
The third area involves autonomous AI systems, which Alsalah described as the next stage of development and an issue receiving significant attention in discussions with banks.
As AI systems become increasingly capable of taking actions independently, a central question is how much autonomy they should be given and at what stages human oversight should remain part of the decision-making process.
