The Ghana Revenue Authority (GRA) is turning to artificial intelligence for plugging long-standing leakages in Customs revenue – after pilot tests showed the technology could lift collections as much as 45 percent by tightening controls on undervaluation and misclassification of imports.
The revenue authority’s push follows findings in the 2026 Budget Statement that exposed widespread abuse of Import Declaration Forms. Between April 2020 and August 2025, more than 525,000 transactions worth about US$83billion passed through the Customs system – but only a small share was linked to actual imports.
The audit found that about US$31billion was transferred abroad without goods entering the country, draining reserves and weakening the cedi. Importers were also found to have under-declared roughly GH¢76billion in goods – denying the state about GH¢11billion of revenue – while some banks processed inflated transfers in breach of central bank limits.
Commissioner-General Anthony Sarpong said this decision followed months of negotiation with the service provider and testing the system through a proof of concept and pilot phase. The outcome, he said, justified both the cost and timing.
“We’ve gone through a very rigorous negotiation with the service providers and we believe that costs the state will be paying for implementing this AI solution will be far lower in terms of value for money,” Sarpong said.
“From the proof of concept and the pilot that we’ve done, if you look at those samples alone, revenues are going up between 40 percent and 45 percent. That is the kind of value we are targetting in deploying this tool for Customs revenue.”
The tax authority plans to deploy an AI-powered trade analytics tool to work alongside the Integrated Customs Management System (ICUMS), which has been the backbone of Customs processing at ports, airports and land borders for more than five years.
Officials say the new layer is designed to address weaknesses that persist despite earlier digital reforms, particularly the heavy reliance on human judgment in classifying goods and assessing duties.
Sarpong said the technology will also address complaints from importers about inconsistent duty assessments. Under the current system, discretion can lead to different outcomes for similar goods.
“If two people bring the same vehicle, because of discretion in the process, one may get a duty which is more favourable than the other,” he said. “When you are using the AI, it tries to present an equal assessment. This ensures parity and accuracy in what determines the duty.”
Speed is another factor driving the reform. GRA estimates that it can take two hours to complete a thorough assessment of a shipment, particularly when classification is complex. With AI support, that process can be reduced to minutes – cutting clearance times while maintaining scrutiny.
“What the AI will do is to give us an advanced quick assessment within five minutes,” Sarpong said. “All the effort our officers go through before they arrive at a position, the AI will support us to do that quickly. The benefit we envisage is efficiency and speed, alongside more accurate determination of duties.”
The planned rollout has raised concerns among importers about whether the software’s cost will be passed on through new fees or levies. Both the revenue authority and finance ministry have sought to calm those fears.
“Government is clear that the cost of implementing this software will not be passed on to importers,” Sarpong said, adding that any new levy would require parliamentary approval and no such policy has been submitted. “There will be no extra costs or levy when we implement this software.”
The service provider, approved by parliament, operates through an investment entity registered in Cyprus – a detail that officials say has already been subjected to legislative scrutiny.
Finance officials frame the initiative as part of a broader effort to raise domestic revenue without increasing tax rates. Deputy Finance Minister Thomas Nyarko Ampem said the AI system, known as Publican Trade Solution, has already been deployed in more than 20 countries and is designed to strengthen Customs controls while facilitating legitimate trade.
Pilot results cited by the ministry show the system detecting misclassification and undervaluation in 18 out of 43 randomly selected transactions. Reviews of five companies linked to those cases led to recovery of about GH¢15million additional revenue.
Beyond revenue, the technology is expected to help Customs keep pace with increasingly sophisticated smuggling tactics as traffickers shift routes and conceal goods to evade detection. By integrating data across borders and agencies, officials say, the system can flag risks that might be missed through manual checks alone.
