Business, Freight News, IT


Home Office uses AI to snare suspect shipments

[ February 13, 2025   //   ]

Home Office Intelligence investigators have developed a new detection approach that queries the HMRC Compliance and Risk Engineering Solutions Team (CREST) for specific types of customs declarations uses other information including Companies House data to intercept suspect freight consignments.

The team of three is identifying 70 potentially suspicious companies each month, and there has been a 30-40% success rate in finding illegal commodities including cocaine, tobacco and synthetic opioids in containers compared to the 7% average for freight searches.

While risk analysis work – including filtering and analysing spreadsheets and desk research – is highly manual, the Home Office is using the Accelerated Capability Environment (ACE) to explore how technology can be used to help automate this.

The aim was to expand the number of companies assessed and maximise the number referred for special inspection.

ACE worked with Home Office Intelligence and Border Force and Vivace supplier Faculty over 12 weeks to develop a rules-based proof of concept risk model that could develop intelligence packages.  By allowing intelligence officers to upload raw HMRC data and combining this with automated retrieval of Companies House data, it reduces the risk analysis process from a week to a few minutes. 

The risk model scores companies on their likelihood of being an importer of contraband according to 11 weighted rules, presentingresults immediately.

An operational trial selected hundreds of new companies for investigation out of which there were a number of successful stops. Initial forecasting shows the tool has the potential to almost double the seizures of contraband from suspicious shipments over the coming months by increasing both volume and hit rate.

Moreover, the data captured will enable machine-learning approaches to be added to increase accuracy even further.

A roadmap and recommendations for how this risk model could be moved into wider operation was developed and further refinement and development is underway.